ABSTRACT. Creaky voice (CrVx) – sometimes labelled vocal fry or glottal fry – is a phonation variant that is auditorily perceived as a rough and low-pitched voice quality. The use of CrVx has previously been documented in many varieties of English, for example, in those spoken in America, in Britain, and in New Zealand. Although large-scale, methodologically rigorous studies on CrVx are relatively scarce, several sociophonetic patterns of CrVx have been suggested, with potential factors including speaker gender, speaker ethnicity and situational context.

In varieties of Australian English, anecdotal observation suggests that CrVx is pervasive and geographically widespread, and that the extent of its use varies considerably across speakers. However, to date, no study has systematically quantified the use of CrVx in varieties of Australian English; who uses it, to what extent, and in what contexts is not known. This study aimed to go some way towards filling this gap. The primary aim of the study was to investigate the prevalence of CrVx in the speech of young male and female speakers from two Australian cities. A secondary aim of the study was to evaluate an automated method for the detection of CrVx intervals centred around the analysis of low and/or irregular f0 known to be a characteristic manifestation of CrVx.

Speech data were obtained from the AusTalk corpus of audio-visual recordings of a socially diverse group of 32 speakers schooled entirely in Australia. Two groups of speakers aged 18-30 years old were included in the study; one group from Perth (n = 16, male = 8, female = 8) and the other from Brisbane (n = 16, male = 8, female = 8). In this initial analysis we used AusTalk’s ‘story re-telling’ task, in which participants were asked to retell a story in their own words immediately after reading it aloud. Drawing on a recent study by Dorreen (2017), intervals of CrVx were identified using Talkin’s (2015) ‘Robust Epoch and Pitch EstimatoR’ (REAPER) analysis toolkit. As discussed by Dorreen (2017), REAPER’s ability to track very low frequency glottal activity (Liberman 2015) provides a basis for identifying intervals of phonation that possess the characteristic low and/or irregular f0 typically associated with intervals of CrVx.

We present results showing the prevalence of CrVx in relation to location, speaker sex, and individual speaker, confirming the anecdotal observation that CrVx manifests as a gradient sociophonetic feature in the performance of English speakers in Australia. Additionally, results show how the prevalence measures are affected by the parameters of the REAPER analysis.

Findings of this preliminary study are considered in light of existing work on CrVx in other varieties of English. Issues arising in the methodology are discussed, with a particular focus on the effectiveness of the REAPER analysis as a means of reliably detecting CrVx. Finally, pointers are given to areas that warrant further investigation.

ABSTRACT. Voice onset time (VOT)—the duration between release of a stop consonant’s closure and the onset of voicing of the following segment—varies as a function of linguistic and extra-linguistic factors. VOT patterning is language- and variety-specific, and covaries with place of articulation (POA), and with identity of the following segment. It is particularly attractive as a sociophonetic variable because it can vary in relation to social factors, including speakers’ ethnic identities and heritage language backgrounds. Variation in stop realisation has been identified previously in relation to speaker ethnicity in Australian English (AusE) by Horvath (1985), and other temporal characteristics of speech have also been shown to vary as a function of Lebanese ethnicity.

To contribute to knowledge on stop realisation and ethnicity in Australia, we present an analysis of VOT in 30 native speakers of AusE with Lebanese heritage. Speakers produced three repetitions of 92 words with word-initial stop consonants /b, d, ɡ, p, t, k/ in stressed syllable-initial position. In presenting a sociophonetic account, our aims are to consider:

1. phonetic factors that condition VOT variability; 2. social factors involved in its patterning; and, 3. interactions between these factors.

We focus on the effects of POA, phonological voicing category, and following vowel. Further, we analyse our results with respect to speakers’ ethnic identities measured by an adapted version of the Ethnic Orientation and Multigroup Ethnic Identity Measure (Revised) (Phinney & Ong, 2007) questionnaires (EOI), and other macro-social categories. We present our analysis using acoustic phonetic methods, and model the data using linear mixed effects regression.

Concentrating on positive VOT only (n = 7,081) we find substantial variability, with VOICED VOT having positive values between 1.5 and 72 ms (M = 16.0 ms), and VOICELESS VOT between 14.3 ms and 178 ms (M = 68.7 ms) (see Figure 1 for e.g. of long-lag VOT). VOT also varies according to POA (p < .05), but, contrary to the findings of classic VOT studies (e.g., Lisker & Abramson, 1964; and studies reviewed in Abramson & Whalen, 2017) which show VOT increasing with more retracted POA, our results show the alveolar place has the longest VOT, followed by the velar and the bilabial, similar to the findings of Docherty (1992) for Southern British English. Results further show VOT varying as a function of EOI, with shorter VOT associated with lower EOI scores. This effect is, however, modulated by speaker sex.

We see evidence of short-lag VOT and long periods of aspiration both in VOICED and VOICELESS stop series of consonants, but overall the categories are maintained by VOT, with VOT further varying according to POA, EOI, and speaker sex.

Thus, this study makes two important contributions to our understanding of sociophonetic variability: it expands our understandings of stop patterning in VOT in Australian English; and it exemplifies the complex ways ethnic identity interacts with other factors like speaker sex in the realisation of fine phonetic detail.

ABSTRACT. Previous studies have described the acquisition of a second dialect by people moving to a new dialect area (e.g. Chambers 1992; Foreman 2003; Tagliamonte & Molfenter 2007). But an increasing number of migrants fall into the category of ‘expatriates’ who are returning home after temporary stays overseas. The patterns of linguistic variation among expatriates returning to their home countries have received no attention at all to date. This paper reports on the first such study within a variationist sociolinguistic framework, a case of what might be called ‘first dialect reacquisition’.

I am presenting results from five of the participants in my study, who lived in the English-speaking expatriate communities of Hong Kong or Singapore before moving to New Zealand between the ages of 5 and 9 years. These children acquired New Zealand English (NZE) as their first dialect from their NZE-speaking parents. When they began interacting with other children, especially at kindergarten and school, studies (e.g. Starks & Bayard 2002) suggest that they would have acquired the variety spoken by their peers as a second dialect. I will be reporting on the phonetic changes made after their move to New Zealand. The participants were recorded at one month and six months after their move and each recording was a semi-structured interview, which lasted approximately one hour.

In this paper, I will be focusing on the changes in realisations of the highly stereotyped NZE short front vowels. Results show considerable variability and overlap in the productions of the KIT, DRESS and TRAP vowels. Nevertheless, Bhattacharrya’s Affinity results show a more marked movement in the DRESS vowel for the youngest speaker. Furthermore, a deeper investigation of the KIT vowel shows a consistent move towards a more NZE-like pattern, and away from the vowel pattern in the original dialect, with the younger speakers patterning differently from the older speakers. The results also show greatest change in realisations of the KIT vowel, followed by the DRESS vowel then the TRAP vowel.

ABSTRACT. Previous research has shown that listeners are fairly accurate in estimating speakers’ age from their speech (e.g. Moyse, 2014). Some research has explored the effect of the speaker’s first language (L1) on age estimation. Nagao and Kewley-Port (2005) presented English and Japanese stimuli to two groups of listeners from these L1s and found that listeners were more accurate at estimating age in the familiar language. Rodrigues and Nagao (2010) extended this line of research to foreign-accented speech by playing English language clips recorded by native speakers of Arabic and English to native English listeners with more and less experience with foreign accents. They found a higher correlation between estimated and chronological age for English speakers than for Arabic speakers and a higher correlation for the more experienced listeners than for the less experienced ones for Arabic speakers but not for English speakers.

The current study sets out to compare age estimation accuracy by first and second language speakers. The audio stimuli were 40 clips of 20 British English speakers and 20 Japanese L1 speakers (age range: 18 - 71) reading a passage in English. Two groups of listeners (36 English and 23 Japanese L1 speakers) were presented with the audio-stimuli and were asked to estimate the speakers’ age.

Statistical analysis shows that male speakers were estimated to be older than females by both English and Japanese listeners and Japanese L1 speakers were estimated to be younger than their English counterparts by English L1 listeners only (Figure 1). This variation can be explained by listener familiarity with the language or accent: English listeners exhibited the highest correlation of 0.64 for English speakers, followed by Japanese listeners performing equally well with English- and Japanese-accented English (0.44 and 0.45), and English listeners listening to Japanese-accented English coming last (0.37). These findings have theoretical implications as they highlight that even such a seemingly universal phenomenon as age may be expressed and perceived differently by people from different language backgrounds and of varying familiarity with languages and accents. This further supports previous studies that show a connection between age and sociolinguistic features, reflecting that age is expressed both physiologically and socio-culturally. The practical implications include our need for awareness of such differences when age estimation occurs in real life (e.g. forensics).

ABSTRACT. Australia has long been known as a “classical immigration country”, whereby the “inflow” of individuals taking up residence in Australia far outweighs the “outflow”, the numbers of individuals that leave Australia for elsewhere (Castles et al. 2014: xiv). This paper presents a study of the adoption of Australian English (AusEng) by one of the most established migrant communities in Melbourne: the Irish, who have a long history of migration to Australia, and whose numbers continue to increase, with a 39% increase in Irish nationals in Australia since 2006 (Australian Dept. Immigration and Border Protection 2016).

Within the variationist sociolinguistic literature, comparatively little is known about the adoption of features of a second or subsequent dialect or variety of the same language—commonly referred to as Second Dialect Acquisition (SDA), although work by Siegel (1997), Markham (1997) and Sigel (2010) has offered important insights. We contribute to the topic with a multilayered research design, with data consisting of (1) sociolinguistic interviews and wordlists; (2) articulatory ultrasound; (3) perception experiments; and (4) written questionnaires pertaining to socio-economic background and makeup of social networks. Attitudinal data elicited in the interviews suggests a range of views towards AusEng, from those that view it as undesirable to those that report having consciously modified their Irish accent to make themselves understood in a professional context. A variety of socio-phonetic, articulatory, discourse-pragmatic and grammatical features will also be examined as part of the broader study. Here we present preliminary results of an acoustic analysis of vowel spaces from five first-generation female participants originating from the Republic of Ireland (Fig. 1), with lengths of residence in Australia ranging from 2 to 9+ years, and a comparator sample of three female AusEng participants, born and raised in Melbourne.

Results show that Irish participants with 9+ years of residence, with almost exclusively Australian social networks, are showing initial signs of movement in the vowel space, such as with a more backed /ʊ/ and a more fronted /ʉ:/ (Fig. 2). Elsewhere in the vowel space, however, participants adhere to general tendencies of IrEng, e.g. by maintaining the merger of hood and Hudd (or FOOT and STRUT) (Ferragne & Pellegrino 2010: 534). This indicates that movement away from IrEng vowels towards the AusEng system (e.g. Cox & Palethorpe 2007) in the first four years of residence may be limited. These data will be further modelled using linear mixed effects regression, and the effect and nature of rhoticity in IrEng with regards to the quality and duration of the vowels will be considered in future phonetic work. As noted by Nycz (2015: 471) socio-phonetic studies of SDA, such as this one, are required for building empirical databases to test and model individuals’ resistance or accommodation to variation and change.

ABSTRACT. Kriol is an English-lexified creole language that has been spoken across large areas of Northern Australian for over a century. Linguistic research on Kriol of Northern Australia commenced half a century ago and extant geographic variation has been widely noted in the literature (e.g. Munro, 2000). Varieties such as Kimberly Kriol, Barunga Kriol, Roper Kriol have been categorised and investigated on their own terms. Variation has been described as predominantly phonetic and lexical (Schultze-Berndt et al., 2013: 242) however research describing and quantifying this variation has not previously been carried out.

This paper analyses one variable, the reflexive particle, which is derived from the English form myself. Ponsonnet (2016) describes the particle and notes phonetic variation among speakers from one community (Beswick/Wugularr) including the forms [miʝɛlp], [miɉʝ͡ɛlp], [mijɛl] and [miɛl] (2016: 303). Ponsonnet’s description incidentally noted there was “not enough published data to tease apart Roper from Barunga Kriol with respect to the features discussed” (2016: 300). In order to begin to analyse variation in Australian Kriol, sociolinguistic interviews were carried out in 2016, surveying over 60 Kriol speakers across nine communities: a sub-area of the Kriol-speaking region typically classified as covering the Barunga and Roper varieties of Kriol. The reflexive particle was identified as one of several dozen variables that vary geographically.

Analysis of the sociolinguistic interviews has been carried out using ELAN, producing a corpus of over 100,000 words covering almost half the project participants. Tagging of the reflexive particle has (to date) produced over 200 tokens, with five main variants categorised. Phonetic analysis is impressionistic, cross-checked by (at least) two transcribers. Findings match Ponsonnet (2016), showing phonetic variation according to the quality of the medial consonant, ranging from realised palatal obstruents to complete elision. Not specifically noted previously is variability in the openness of the final vowel, producing forms that range from [ɛl] to [al] in the final syllable.

Early results show that forms with the open final vowel are associated with Eastern Roper Kriol speakers, a pattern also seen in other variables (e.g. ‘there’ [dɛ:], [ʝa:]). This may be attributable to three-vowel systems found in several substrate languages of the Roper region. Regarding the medial consonant, it appears that speakers from the easternmost communities surveyed are more likely to produce the fully realised obstruent, whereas elided and lenited forms are more common in westerly communities in the region surveyed. These results build upon the analysis of the reflexive particle given by Ponsonnet (2016) and provide a first step in building an overall understanding of geographic variation in Australian Kriol.

ABSTRACT. Variation in agreement with plural existential there-constructions, as shown in (1), is a feature of all varieties of spoken English (Chambers 2004). However, the social and linguistic factors conditioning this variation have proven inconsistent across studies, even in data from the same variety (Walker 2007, 2014).

(1) a. We had to hang our food because there were bears. (Speaker 4: 774) b. It helps being in the city where there is other kids around. (Speaker 70: 193) c. There’s black bears, I believe there’s brown bears. (Speaker 6: 59)

At least some of this inconsistency stems from the ‘small n problem’ of grammatical variation: grammatical variables occur infrequently in speech, which results in much smaller datasets and fewer tokens per speaker. This not only reduces the reliability of tests of statistical significance but also lessens our confidence in social effects and the extent to which speakers share linguistic conditioning within groups.

This paper investigates these problems by analysing the same dataset from different perspectives. Using a corpus of Toronto English (Hoffman & Walker 2010), I extracted 1,039 tokens of plural existential there-constructions from sociolinguistic interviews with 69 speakers stratified according to ethnic background, generation and sex. Each token was coded for a series of linguistic factor groups as well as the above social factors. Two methods of analysis were used: mixed-effects models (MEM), a type of logistic regression in which the speaker is modelled as a random effect and all other factors as fixed effects, and conditional inference trees (CIT), which are constructed by recursive partitioning according to factors that significantly divide the data. These two methods are conceptually different: one involves hypothesis testing and the other is a form of data exploration.

Following previous studies (Walker 2007, 2014), I distinguish between singular agreement (1b) and there’s (1c) as separate variable processes. MEMs select several linguistic factors (polarity, determiner type and intervening material) as significant for both variants, but social factors (generation, ethnicity) only for there’s. However, the high degree of overlap between social and linguistic factors vitiates the use of MEMs, and for some speakers the number of tokens is too low to perform analysis of linguistic factors within each group. The CITs show an overriding effect of temporal reference for both variants but reveal secondary linguistic effects for singular agreement and social effects for there’s. Including speaker as a factor in the CITs does not reduce the social effects, and all of the major effects are confirmed through generating random forests of trees.

An important finding of this comparison of methods is that they do not contradict each other in revealing the linguistic and social factors conditioning variable agreement in existentials. However, the use of CITs provides an additional perspective in assessing the relative importance of factors when the data are sparse (small n) but the effects are significant (big p). The results of this study provide support for using different statistical techniques in order to understand the nature of grammatical variation.

ABSTRACT. How can we access the cognitive processes governing variation? While surveys and experimental methods are commonly used, such data cannot provide access to the fine-grained patterning in spontaneous speech. Priming – the tendency for speakers to repeat previously mentioned structures – has been used as a window on variable grammar (cf. Tamminga, 2016). Here, priming is used as a measure of speaker associations between forms for the second person singular (2sg) in Chilean Spanish.

Chilean Spanish is characterised by variation between tuteo: the 2sg familiar pronoun tú and corresponding verb forms as in (1) below; and voseo: the 2sg pronoun vos with corresponding verb forms as in (2)). In Chile both tú and vos co-exist, and the two paradigms often mix, as in (3).

(1) No po si tú tienes razón. ‘No you-TÚ are-TÚ right.’ (2) Vos cachái que ese estadio lo hicieron, ‘You-VOS know-VOS they made that stadium,’ (3) ... Pero tú tenís que trabajar po. ‘But you-TÚ have-VOS to work.’ (4) Qué [Ø] tenís que comprar. ‘What do youØ have-VOS to buy.’

While the pronoun vos is stigmatised and (virtually) absent in spontaneous data, voseo verb forms are predominant. Nonetheless, Chileans routinely report using tuteo and avoiding voseo (e.g. Stevenson, 2007: 165). In survey data (Bishop and Michnowicz, 2010: 419), speakers (N=80) self-reported using voseo at most 25% of the time. However, in a corpus of conversational Chilean Spanish (105,000 words, 35 speakers) the voseo rate is 80% (819/1,013). The divergence between self-reports and real rates of usage might suggest that speakers do not recognise the voseo forms as belonging to a distinct paradigm from tuteo. By using priming as a measure of speaker associations between constructions (Travis et al., 2017: 13), it is possible to apply an empirical test of this.

Priming across verb forms is evident in the data, indicating that speakers do attend to the difference between the paradigms: after a preceding voseo, the rate of voseo rises from 80% overall to 92% (245/265), and the rate of tuteo skyrockets from 20% overall to 72% (43/60) following a preceding tuteo. Further analysis shows an association between tú and tuteo – speakers avoid using the tú pronoun with voseo verb forms (as in (3), and instead use an unexpressed subject, as in (4)). This effect, however, disappears in the context of a preceding voseo: the voseo verb form is preferred independently of the presence of a pronoun. This suggests that the stronger cognitive association between verb forms (voseo  voseo, and tuteo  tuteo) overrides the impact of the pronoun. Analysis of actual usage provides evidence that speakers do in fact distinguish voseo verb forms from tuteo, contradicting the self-reports. Moreover, the priming shows that, in spontaneous speech, the verb forms actually play a more important role than the reportedly more salient pronouns (tú and vos).

ABSTRACT. General extenders (GEs) have become a showcase variable for the application of the variationist method to discourse features. Quantitative explorations of GEs in English spoken in the UK, the US and Canada reveal a rapidly changing system, with many similarities across varieties of English, including for Australian English. To date, these changes have been considered from the perspective of apparent-time. Here, we combine real- and apparent-time analyses to GEs in Australian English, to better understand the nature of the changes and how they spread through the system and the society.

The data come from the Sydney Speaks project (led by Travis, 2016-2021), which brings together three corpora. Recordings made from 1977-1988 provide apparent-time comparisons across three age cohorts: older speakers, born circa 1900 (NSW Bicentennial Oral History Project 1987); adults, born 1930s; and teenagers, born 1960s (Horvath 1985). The Sydney Speaks 2010s corpus, under compilation, adds 30 years in real time, through recordings with older (born 1960s) and younger adults (born 1990s). Here, we analyse GEs occurring in approximately 600,000 words of spontaneous speech drawn from these three corpora, comprising close to 100 speakers of Anglo heritage, stratified according to age, gender and socio-economic status.

Analyses of 2,000 GEs exhaustively extracted from the data allow us to chart change in both the repertoire of forms in the system and in the behaviour of individual forms; we focus on the four most frequent forms—or something, and things, and that and and stuff—illustrated in examples (1-4) below.

On the whole, the real-time 2010s comparisons confirm the predictions deriving from the 1970s apparent-time study. We observe the decline of some forms (e.g., or something, and things), the introduction of others (e.g., and stuff), and a change in shape of yet others (e.g., and that moves towards (and that) kind of thing). However, these changes are not accompanied by systematic change in the semantics of the forms, such as semantic bleaching, operationalized as a move away from the canonical use of implying extension from some general set (cf., Cheshire 2007:174). Nor is there evidence of systematic phonetic reduction associated with the changes in the system; rather, for and that a higher rate of and reduction is associated with those who use the form the most (namely, 1970s teenagers and 2010s adults).

Thus, our findings suggest a GE system in Sydney English that is undergoing lexical replacement without accompanying grammaticization (parallel to the proposal put forward for Toronto by Tagliamonte & Denis 2010). This research adds to the growing body of literature that investigates the relationship between discourse-pragmatic variables and their social context, demonstrating a high degree of systematicity within complex variability.

1. Pam she'll probably look at me as though I'm a nut or something. [AAFU-303: 18:26] 2. Carol and you're in swimming clubs and things. [AAFU-259: 09:17] 3. Mitchell I was so sore and that. [ATMW-128: 40:28] 4. Hayley I found that comforting to like identify with and stuff. [AYFU-014: 47:11]