V: Future Directions for Panel Research

What's the Point of Panel Studies?

Suzanne Evans Wagner

This volume might come as a surprise to some observers. A collection of panel studies of language variation and change was published here in the Routledge Studies in Language Change series only a few years ago (Wagner and Buchstaller 2018). Why do we need another one so soon? After all, as the 2018 volume amply discussed, panel studies2 are beset with methodological and analytical problems, including the Small n Effect and Gap Effect (Cukor-Avila and Bailey 2018), as well as confounds with style-shifting (Wagner and Tagliamonte 2018), physiological aging, and second dialect acquisition (Reubold and Harrington 2018; Kohn and Farrington 2018). So is there any point in doing them? The short answer is: Yes. Panel studies are absolutely worth undertaking and worth publishing regularly, and for (at least) three good reasons.

First, the present volume, and others like it that will appear in the future, provide a venue for studies of the typically fragmentary, sparse data that often characterize panel studies and thus cannot be subjected to the complex statistical techniques now expected from many peer- reviewed journals in linguistics. Sociolinguistics and its allied fields have made massive strides in the last decade with respect to the sophistication and variety of its statistical methods. One need only look at any quantitatively oriented journal publishing work on language change to find terms that were rarely present ten years ago, such as tensor product smooths, Generalized Additive Models, random slopes, and conditional inference trees. Yet many panel studies are necessarily limited to small numbers of speakers (often just one or two) and to small amounts of speech from each individual, leading to a preponderance of studies in which the analyst can do little more than tally up some frequencies and calculate a percentage or maybe conduct a chi-square test. As a result, the numbers and chi-squares generated in small panel studies are even more vulnerable than they were in the past to “yes, but...” questions: “Yes, but what about the fact that the interviewer was different at each time point?”; “Yes, but what about the preponderance of lexical item X in the data?”; “Yes, but what about the likely interaction of social class and gender?”; “Yes, but these data are uncontrolled for age at Time 1” and so on. Yet even though digital academic publishing is increasingly creating space for short papers, null results and preliminary or speculative findings, there is still a vital role for books like this one.3 As Labov (1994: 76) observed, a panel study might have “a reduced sample, perhaps too small for statistical significance, but [it is] nonetheless extremely valuable for the interpretation of the original observations”. Put another way, good analysts can draw meaningful conclusions from sparse data, and we should publish those conclusions.

Second, it is precisely because panel studies are often limited that we need to undertake more of them and to publish them. This will help them to garner the attention and scrutiny they require and, in doing so, will increment our overall knowledge of the interaction between language change in the individual versus the community. Variationist sociolinguistics, from its inception, has always proceeded by building up generalizations from case studies. One of the most cited studies in the field is Labov’s (1990) “The intersection of sex and social class in the course of linguistic change”. This paper distills two “principles”, or generalizations, about language change from 57 studies of 16 languages and at least 45 speech communities published 1905-1990. This approach benefits panel research as well. Panel studies come in different shapes and sizes and have different strengths and weaknesses. Occasionally they can be directly compared within a single analysis, as Sali Tagliamonte and I have been able to do with our complementary “Clara” and South Philadelphia studies (Wagner and Tagliamonte

2018). And if we are to regularly incorporate historical written sources into our models of language change across the lifespan, as this volume does (Hernandez-Campoy 2021; Standing and Petre 2021), it is even more imperative that we tolerate - while continuing to interrogate - their inevitable data skews and gaps in interpretation. The challenges posed by panel studies are not insurmountable, and certainly cannot be overcome unless we are actively sharing potential solutions. For example, Kohn and Farrington (2018) provide practical suggestions for handling so- ciophonetic data produced before, during and after participants go through puberty; while Buchstaller, Krause-Lerche and Mechler (2021) propose a triangulation technique for dealing with the Small n Effect.

Third, the variationist study of individual lifespan linguistic change, while certainly not new (see, e.g., Prince 1987), has always been of justifiably secondary concern to variationist sociolinguists, whose focus has been on the collective community grammar. Besides, prior to the widespread availability of longitudinal digitized corpora, panel studies were undertaken only by those with the sheer persistence needed to conduct them or with the good fortune to stumble upon relevant data sources. Consequently, we are now seeing an academic goldrush, as linguists seize on new datasets or re-examine old ones that had perhaps been stored away on magnetic tapes in a lab cupboard or at a radio station archive (Cieri and Yaeger-Dror 2018). The present volume is a reflection of the exponential growth that is occurring in studies of lifespan language change.

One advantage to panel studies being a still-nascent area in quantitative sociolinguistics is that their practitioners are compelled to seek out insights from adjacent (sub)fields in which they are better established. This has led to a cross-disciplinary exchange of information that can surely only accelerate the science. For example, in their со-edited special issue on language and aging in Linguistics Vanguard, Gerstenberg and Lindholm (2019) take care to bring together researchers trained in var- iationist sociolinguistics, psychology, medical sciences, and discourse analysis. Wagner and Buchstaller’s (2018) volume similarly gathers lines of inquiry from across disciplines and subfields including education, second language acquisition, corpus linguistics, linguistic anthropology, and phonetics. Yet at the same time, the wide scatter across disciplinary boundaries requires scholars of lifespan linguistic change to do a lot of time-consuming bibliographic detective work, and to keep up with advances in multiple (sub)fields. Edited collections and special issues are therefore of particular value.

Edited collections also serve as a cross-sectional snapshot of the state of the field at the moment of publication and a commentary on where the field is going next. In what follows, I share some of the future directions for panel studies that I think can be extrapolated from the present volume.

< Prev   CONTENTS   Source   Next >