Variationist presentations

Wednesday 8 February

Plenary

Form-function (a)symmetry and the pursuit of categoricity

Shana Poplack, University of Ottawa

Because the doctrine of form‐function symmetry is so firmly entrenched in linguistic thought, the existence of the linguistic variable beyond the phonological is often dismissed. In this talk I examine the response to morphosyntactic variability over time by tracking its treatment in a massive corpus of prescriptive grammars dating from the 16th century through the present, and relating it to current formal approaches. Analysis shows that although variant forms have been recognized since the earliest times, only rarely have they been acknowledged as variant expressions of the same meaning or grammatical function. Instead three major strategies are marshaled to factor variability out, when it isn’t ignored altogether: assigning each variant a specific linguistic context, matching each variant with a dedicated meaning, or (when all else fails) associating each variant with a distinct type of speaker or register. Remarkably, however, there has been little consensus, whether over time or across grammarians (and linguists!), over which elements to associatewith which variant. This suggests that the aim is of these strategies is not so much to prescribe or even describe, but simply to imbue each form with a privative context of occurrence, whatever it may be, so long as it is distinct from that of its counterpart(s). Attributing distinct roles to each variant restores the desired isomorphic relation between function and form, while implicitly rejecting the possibility of bona fide grammatical variability. Systematic confrontation with the data of spontaneous speech fails to validate virtually all of these treatments, however, revealing robust variability subject to regular conditioning instead. I explore how the enduring legacy of this position, encapsulated in the doctrine of form‐function symmetry, continues to mold not only prescriptive, but also many formal linguistic treatments of variability, contributing to the growing gulf between prescription, description and actual usage.

Shana PoplackShana Poplack, C.M., FRSC, is Distinguished University Professor and Canada Research Chair in Linguistics at the University of Ottawa, and director of the Sociolinguistics Laboratory there. She is a sociolinguist who studies language use in everyday situations, with a particular focus on bilingual and minority-language contexts. Her work applies theoretical and methodological insights gained from the study of linguistic variation and change to a variety of fields of interest to linguists, including constraints on bilingual language mixing, language contact and grammatical convergence, the genesis of African American Vernacular English, language ideology, normative prescription and praxis, and the role of the school in impeding linguistic change. This research is characterized by the collection and scientific analysis of large bodies of natural speech data, many of which are one of a kind. They are housed at one of the most dynamic and productive sociolinguistics laboratories in the world. The lab has trained scores of students and associates at all levels, and its computerized spoken-language resources have served as data for hundreds of publications, presentations, theses and dissertations.


Session 1: Variationist typology


The locus of cross-linguistic similarity and difference

Catherine Travis, Australian National University

Language classifications are often based on the presence or absence of a particular feature, or set of features. One such example is the distinction between “null-” and “non-null subject languages” (e.g. Spanish, Italian vs. English). Here, in what we term Variationist Typology, we bring new insights into cross-linguistic tendencies through comparison of the variable structure internal to different languages. The locus of comparisons is the set of probabilistic constraints on the variation as well as the delimitation of the variable context within which the probabilistic constraints are operative. We demonstrate through consideration of the similarities and differences in subject pronoun expression in English and Spanish.

Catherine TravisCatherine Travis is Professor of Modern European Languages in the School of Literature, Languages and Linguistics at the ANU. Catherine’s work is focused on language variation and change, which she is currently exploring through two main projects. One involves testing the convergence via code-switching hypothesis, through examination of the Spanish spoken in a bilingual community in New Mexico, USA (NMSEB project). The other concerns the study of social variation in English spoken in Australia, through a real and apparent time study of spontaneous speech of diverse social groups (including migrant communities) over the past 40 years (Sydney Speaks).

Lenition: Sociolinguistic and phonological typology

John Mansfield, University of Melbourne

Following Torres Cacoullos and Travis’ (to appear) “Variationist Typology”, this talk proposes a next step for the endeavour: Can we compare a variable feature across languages, comparing both the linguistic constraints, and the social constraints? I present here an exploratory study, taking obstruent lenition as a variable typological feature that has been documented in languages including Yolngu, Pitjantjatjara, Murrinhpatha, Italian, Spanish and English. In this dataset we find that lenition has quite different linguistic constraints in different languages – in particular, being potentiated in different loci of prosodic structure. However, the social constraints on lenition are similar across the languages studied: it is favoured by solidary, low-prestige, and youth speech.

John MansfieldJohn Mansfield researches how the social uses of language are manifest in diverse language types, especially the Aboriginal languages of northern Australia. Language use reflects the social role of speakers, the stances they take in verbal interaction, and the unfolding of language change over time. Different language types offer speakers very different repertoires for expressing linguistic identity.

A system in motion: Morpho-syntactically conditioned phonetic variation in Hawai'i Creole

James Grama, Australian National Unviersity

One criticism of variationist research is that it tends to focus on single variables, while relatively little research addresses whether multiple variables pattern similarly within individuals (see Guy 2013). The simultaneous treatment of multiple variables across different levels of linguistic representation is reminiscent of the importance of assessing intra-linguistic behaviour in variable contexts, as identified by Torres Cacoullos and Travis (To Appear). In the present talk, I discuss how investigating multiple variables simultaneously in a language is useful in capturing co-dependent variation in a single language—Hawaiʻi Creole—where acoustic phonetic vocalic variation is mediated by the rate of use of Hawaiʻi Creole morpho-syntactic items.

James GramaJames Grama is a Postdoctoral Fellow at the ANU whose work is focused on language variation and change, specifically with respect to acoustic phonetic and phonological variables. Much of his research investigates this type of variation in under-studied English dialects and English-based varieties, and he has a particular interest Hawaiʻi Creole. Currently, he is working on quantifying, describing, and contextualizing the social variation in the English spoken in Australia across social class, gender, ethnicity, and social network across both real and apparent time as a member of the Sydney Speaks project.


Session 2: Wellsprings session


Session chair and opening remarks

Nick Evans, Australian National University

Nick EvansNicholas Evans is a typologist and anthroplogical linguist specialising in the languages of Australia (Kayardild, Bininj Gun-wok, Dalabon, Iwaidja) and Southern New Guinea (Nen). He has long been intrigued by the typological and analytical problems posed by polysynthetic languages, issues he has grappled with in his grammar of Bininj Gun-wok (Evans 2003), his coedited book Problems of Polysynthesis (with H-J Sasse, 2002) and his dictionary of Dalabon (Evans et al 2004). His other major contributions include work synthesising linguistic and archaeological evidence (McConvell & Evans 1997), a grammar of Kayardild (Evans 1995), a widely-translated book on the the scientific and humanistic value of endangered languages (Evans 2010), and work arguing for a linguistics built on language diversity as a foundational premise (Evans & Levinson 2009). Based at the Australian National University, he directs CoEDL, the ARC Centre of Excellence for the Dynamics of Language.

Linking Sociolinguistic Variationism and Language Diversity Studies

Eri Kashima, Australian National University

The relationship between the study of language variation on the one hand, and change on the other, is well-articulated throughout the literature. What is less explicit is the relationship between language variation, and linguistic diversity. In this talk I will make the case for employing synchronic studies of sociolinguistic variation to better account for the linguistic diversity we see globally. This will be done by emphasising the role of linguistic divergence in language change. The overall argument is that studying sociolinguistic variation and change can contribute to our understandings of social factors in language divergence processes, which in turn can help make sense of language diversity.

Eri KashimaEri is a PhD candidate at the Australian National University, who forms part of the Southern New Guinea contingent of the Wellsprings of Linguistic Diversity Project. She has completed two honours theses at the University of Melbourne; one in Anthropology (2007), and the other in Linguistics (2013). Her research interests lie in language contact and change phenomena, multilingualism, and sociolinguistic variation.

How far does language change during the lifetime?  Fitting S-Shaped Curves to Apparent-Time Variation in Southern New Guinea

Mark Ellison, Australian National University

For more than half a decade, linguists have described the progress of language change with s-shaped curves (Osgood and Sebeok 1954). The mathematical description of the curves has varied from Normal cumulative distribution functions to logistic functions, but the shapes are similar. The s-shaped curves have been given a number of functional explanations (see Kroch 1989, Blythe and Croft 2012).

In this talk I present a Bayesian approach to fitting s-shaped curves to variation data augmented with speaker year of birth. As with other apparent change modelling, we hypothesise that speakers reflect not only the current distribution of language use, but distributions they encountered earlier in their lives as well. If a change from form A to form B for a particular construction is in progress, then older speakers are - under this assumption - more likely to use the conservative form A than their younger fellows.

If the apparent-time differences reflect a historical change, we should see an s-shaped curve in the propensity to use the new form rather than the older form as a function of year of birth. A simple family of models is constructed whereby year of birth is mapped via a normalised hyperbolic tan function to the likelihood of using the innovation. The family has two parameters: the maximum rate of change, and the time at which this occurs. Bayes' theorem permits the calculation of a distribution of parameter settings for any given data set.

I examine two distinct case studies. The first is the grammaticalisation of the copula in a prominence marker construction in Nen (nqn). The second is the progress of h- deletion in Nambo. I discuss the dating of the change based on the data we have, and how our understanding of speaker plasticity impacts on the dates we infer.

Mark EllisonMark Ellison studied Pure Mathematics at the University of Sydney, but even then was interested in language change and reconstruction. At the University of Western Australia, this interest evolved into a PhD on machine learning and phonology. The focus on phonology lead to 3 years research work in Computational Phonology at the University of Edinburgh, and subsequent lecturing in Cognitive Science there. Mark left academia in 1998 to learn Polish and work in IT. Recently at the University of Western Australia, he's been using Experiment Semiotics to model language origins, but is now excited to join this project at ANU.

Variation in expression of human referents across languages and in kinship terms in Matukar Panau

Danielle Barth, Australian National University

I present results from a cross-linguistic study of 12 languages’ expression of human referents in the Family Problems Picture Task (San Roque et al., 2012). Are people described minimally, by their role in society or are they put in relationships to one another? I then focus on the patterns of (possessed) kinship data seen in Matukar Panau, where recent developments mean it no longer behaves like a normal Oceanic language. New terms coming into the system have shifted the border between alienable and inalienable possession in the language. Finally, I discuss which variables found in the Matukar Panau data may scale up for a variationist cross-linguistic study.

Danielle BarthDanielle Barth received her PhD at the University of Oregon, focusing on the interface between syntax, phonetics and information theory. She uses empirical data drawn from corpora, experiments and descriptive fieldwork. Her current primary work is on describing variation in Matukar Panau (Oceanic, Papua New Guinea), and building a cross-linguistic corpus with a team of researchers to investigate features of social cognition.

Her primary areas of interest are complex predicates, possession and nominalization, probabilistic reduction, kinship terminology and grammaticalisation. She is interested in various quantitative methodologies and particularly how we can apply them to smaller data from non-majority languages.


Session 3: Variation and change across an entire language


Visualising the effect of multiple variables

Miriam Meyerhoff, University of Wellington & Steffen Klaere, University of Auckland

We undertake a detailed analysis of a sample of over 10,000 utterances from 18 speakers in a corpus of Bequia English (St Vincent and the Grenadines) and apply constrained cluster analysis to discern patterns that identify the linguistic signatures for different villages and to see how individuals pattern in relation to the rest of their village. An analysis of multiple variables provides a clearer picture of intergroup differences than any one variable does. We suggest that this approach promises a better understanding of the mysterious mechanisms by which variation between individuals scales up to variation between groups.

Miriam MeyerhoffMiriam Meyerhoff is Professor of Linguistics at Victoria University of Wellington. Most of her research studies variation in situations of language contact and she’s done fieldwork in Vanuatu, St Vincent the Grenadines, as well as studying patterns of variation arising from urban migration. The expressive component of structured variation complements an interest in the expressive dimensions of gendered language use.

Steffen KlaereSteffen Klaere is Senior Lecturer in Statistics at the University of Auckland. He has investigated probabilistic models for phylogenetic inference and their relevance in application. More recently, he is investigating statistical ecosystem models to study microbial communities and their interaction with other organisms. He is always amazed at the general similarity of research questions in different fields from the statistical point of view, and the essential differences in the finer detail of the actual subject studied.

Applying population genetics to multiple linguistic variables

Xia Hua, Australian National University

Population genetics describes how allele frequencies change in populations over generations. The analogies between changes in allele frequencies and changes in language usages stem from the fact that the transmission of an allele from parent to offspring and a person reproducing a token that he/she has heard are both replication processes. The Wright-Fisher model is often used to test whether alleles in the next generation are random samples from gametes produced by the previous generation. We can adapt this model to test whether different variants of multiple linguistic variables have the same probability to adopt over generations.

Xia HuaXia Hua is a postdoctoral scientist at the Australian National University. Her research focuses on fundamental questions in evolutionary biology, particularly the processes of speciation and diversification. Her work spans from statistical analysis of patterns of DNA evolution to developing new computation methods in comparative biology to formulating theoretical models of evolutionary change to empirical studies of macroevolutionary patterns. She has been collaborating on adapting methods in evolutionary biology to understand the evolutionary process of languages, from micro-scale, such as changes in language usages over generations, to macro-scale, such as factors on the rate of diversification in languages.

Methods for gethering data for multiple variable analysis

Felicity Meakins, University of Queensland

The application of population genetics methods such as the Wright-Fisher model to language data means we are now able to map variation and change across an entire language, rather than just a single linguistic feature. These methods are difficult to apply to existing corpora because they generally have a limited number of speakers recorded across diverse contexts which means that the data is not comparable. To study the formation of Gurindji Kriol, we examined 131 language variants from 78 speakers across three generations from an 80 hour (57,179 clause) annotated corpus. The corpus contained data from picture description tasks, such as Frog Stories, which means that each speaker had a number of opportunities to express the linguistic feature of interest using a Kriol, Gurindji or an Innovative variant. The result is a binary coded database with 22,698 data points which can be used to model the development of Gurindji Kriol across three generations of Gurindji people.

Felicity MeakinsFelicity Meakins is a Senior Lecturer at the University of Queensland. She is a field linguist who specialises in the documentation of Australian Indigenous languages in the Victoria River District of the Northern Territory and the effect of English on Indigenous languages. She has worked as a community linguist as well as an academic over the past 16 years, facilitating language revitalisation programs, consulting on Native Title claims and conducting research into Indigenous languages. She has compiled a number of dictionaries, text collections and grammars of traditional Indigenous languages and has written numerous papers on language change in Australia.


Session 4: Large discussion about variation


Four breakout groups

Nick Thieberger, Nick Evans, Catherine Travis, Felicity Meakins


  • Australian Government
  • The University of Queensland
  • Australian National University
  • The University of Melbourne
  • Western Sydney University