Corpus Approaches to Discourse

A Critical Review

Charlotte Taylor, Anna Marchi, Charlotte Taylor, Anna Marchi

• Publisher: Routledge
• Year: 2018
• Print ISBN: 9781138895782
• eBook ISBN: 9781351716062
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

Corpus linguistics has now come of age and Corpus Approaches to Discourse equips students with the means to question, defend and refine the methodology. Looking at corpus linguistics in discourse research from a critical perspective, this volume is a call for greater reflexivity in the field. The chapters, each written by leading authorities, contain an overview of an emerging area and a case-study, presenting practical advice alongside theoretical reflection. Carefully structured with an introduction by the editors and a conclusion by leading researcher, Paul Baker, this is key reading for advanced students and researchers of corpus linguistics and discourse analysis.

1

Introduction

Partiality and reflexivity

Anna Marchi and Charlotte Taylor

1.1 Aims

Corpus linguistics and the analysis of discourse are described by Sinclair (2004: 11) as ‘the twin pillars of language research’ and, in his words, ‘what unites them is […] they both encourage the formulation of radically new hypotheses [and] the dimensions of pattern that they deal with are, on the whole, larger than linguistics is accustomed to.’ Thus we have a long-standing and natural synergy between the two, based on their individual goals and foundations. This combination has become increasingly popular over the past decade in which we have seen a dedicated conference series emerge (Bologna 2012; Siena 2016; Lancaster 2018) and in 2017 the Journal of Corpora and Discourse Studies was established. The increased prominence of the combination may also be seen if we look at articles published in the journals Corpora and International Journal of Corpus Linguistics; in the period from 2015 to the time of writing, over 40 per cent referred to discourse in the title, keywords or abstract. This interest is also matched by scholars coming from a primarily (critical) discourse studies background who are moving into corpus analysis. Thus we have a situation where corpus & discourse methodologies are being combined and, while some researchers are doing both from the beginning of their career, many are incorporating one or the other approaches into their existing research repertoire. As Mautner (2016: 171) points out: ‘as a junior researcher you are likely to be socialized into the one methodology or the other, but rarely into both.’ Therefore, this book is aimed at a large family of scholars who use corpora to study how social realities are constructed, represented and transmitted linguistically, in other words, people who work with corpora ‘to make sense of the ways that language is used in the construction of discourses’ (Baker 2006: 1).

There have been a number of publications explicitly addressing this relatively new audience (such as Baker 2006; Charles, Hunston & Pecorari 2009; Partington, Duguid & Taylor 2013; Baker & McEnery 2015, inter alia). However, to date there has been relatively little dedicated space for reflective or reflexive work on what the combination of corpus linguistics and (critical) discourse analysis can achieve and what kinds of theoretical assumptions are embedded into the methodological choices (with some notable exceptions, discussed below). What we want to argue here is that as corpus & discourse research has come of age, the time has come to pause and reflect on what it is we do. As the editors and contributors are practitioners of this methodological combination, we aim to look at our practices critically and reflexively. This is not intended as an assault on the exciting work which is emerging, but a recognition of the maturity of the methodology which is now robust enough to withstand, and indeed demand, close scrutiny.

One of the key assumptions regarding added values in corpus & discourse work is that corpus linguistics can provide greater objectivity to (critical) discourse analysis. ¹ However, at least in part, this assumption is based on the misapprehension that the latter is purely qualitative and the former purely quantitative. And this in turn seems to be based on a misreading of what each of these terms involves, as we discuss in Section 1.2. The role of discourse analysis cannot simply be reduced to close reading. If we recognise, instead, that the various toolkits employed within (critical) discourse analysis offer a highly structured set of resources for classifying and interpreting language features, which gain salience against a pattern of (un)usuality, then it is easy to see that discourse analysis inherently has quantitative potential. No pattern can be identified unless its components are quantifiable. Similarly, as corpus linguistics is the use of corpus tools by researchers who themselves determine which linguistic questions are worth posing, and who interpret the findings at each stage, it too necessarily invokes a qualitative dimension. In this regard, another key similarity between corpus linguistics and discourse studies is that neither is a standardised set of approaches to data and therefore any combination involves selection from a range of resources. Just as McEnery and Hardie argue that corpus linguistics ‘is not a monolithic, consensually agreed set of methods and procedures for the exploration of language’ (2012: 1), so van Dijk (2013, no page) asserts that ‘contrary to popular belief and the unfortunate claims of many papers submitted to discourse journals, CDA is not a method of critical discourse analysis’ because the methodologies employed are so diverse.

Mautner (2016: 171–176) identifies six potential areas of concern in combining corpus linguistics and critical discourse analysis: skills gap and lack of standardisation, institutional barriers, resisting temptation in data collection, decontextualised data, language innovation and epistemological issues. It is principally to this last point which we dedicate our attention. This volume systematically questions assumptions of guaranteed ‘objectivity’ as a result of the combination of corpus linguistics and (critical) discourse analysis and addresses the theme of partiality in this research. By partiality (discussed further in Section 1.3) we mean both the potential incompleteness of research (for instance, focussing on difference at the expense of similarity) and potential bias or influence (for example, the influence of the individual researcher, or of individual research techniques).

We aim to cast light on the ways in which methodological choices affect the outcome of the research and to use case studies to show how researchers may enhance the completeness (‘non-partiality’) of their analysis. Thus, we may summarise the main objectives of the work collected in this volume as:

1 Greater awareness of the implications of methodological choices.

2 Exploration of neglected or under-studied areas.

3 Improved toolkit for carrying out research combining corpus linguistics and discourse analysis.

In the remainder of this introductory chapter, we discuss the concept of the qualitative/quantitative relationship which underpins so much discussion of the combination of corpus & discourse analysis and the different kinds of partiality that may affect the direction and interpretation of our research.

1.2 Qualitative and quantitative

A fantasy dialogue circulates amongst social scientists staging an exchange between the American sociologists Charles Wright Mills, heir of the critical school, and Paul Felix Lazarsfeld, founder of empirical sociology, taking the famous incipit of Wright Mill’s Sociological Imagination:

C. W. M.: ‘Nowadays men often feel that their private lives are a series of traps.’

P. F. L.: ‘How many men, which men, how long have they felt that way, which aspects of their private lives bother them, when do they feel free rather than trapped, what kinds of traps do they experience, etc.?’

(Elcock 1976: 13)

Quantitative and qualitative purists have long been in opposition, they have conceived quantitative and qualitative research as incommensurable paradigms, and their results as incomparable (Howe 1988). Paradigms are ‘universally recognised scientific achievements that for a time provide model problems and solutions to a community of practitioners’ (Kuhn 1970: viii); in time a paradigm exhausts its explanatory power and it is surpassed by a new paradigm. By definition, then, paradigms are in competition with one another and conceptualising qualitative and quantitative research in these terms makes the two appear mutually exclusive: We can shift from one to the other but we cannot merge them (Guba 1990). In the context of the 30-year ‘paradigm wars’ (see Tashakkori & Teddlie 1998), mixed methods emerged as a ‘third way’ (Johnson & Onwuegbuzie 2004), following the idea that research should combine methodologies in order to suit the matter under investigation.

The principle behind a non-paradigmatic view of methodologies is that methods should derive from the research question and follow its needs. Wright Mills (1959), who famously criticised the twin evils of ‘grand theory’ and ‘abstracted empiricism’, warns that methods should suit the social problem and not the other way around. Qualitative and quantitative are labels we attach to what we do and methods should be tools, not schools. Therefore, our conceptualisations and identifications should be instrumental rather than deterministic:

I have never presented myself as a single-minded advocate either of ‘qualitative’ or ‘quantitative’ methods. I have certainly stressed the need for adopting methods appropriate to research questions.

(Oakley 1999: 252)

It is in this instrumental perspective that the discussion about combining the quantitative rigour of corpus linguistics with the social perspective of qualitative approaches to discourse analysis finds its frame.

In her seminal 1995 paper Hardt-Mautner speaks of the integration of corpus linguistics and critical discourse analysis in terms of the selection of the best tools available from the two approaches. She discusses the benefits of integrating them, presenting a win–win situation where the methodological mix allows the researcher to overcome the limitations of the individual methods, while benefiting from their respective qualities. On one side we have the ‘greater precision and richness’ of qualitative analysis; on the other we are also granted the ‘statistically reliable and generalizable results’ (McEnery & Wilson 1996: 77) of quantitative studies. Corpus linguists often complain that discourse analysts rarely use quantitative tools to describe the extent to which their findings are generalisable. Discourse analysts, on the other hand, criticise corpus linguists for mainly focusing on individual words and lacking insight into the various dimensions of discourse structure and into the extralinguistic knowledge needed to get to the ‘hidden story’ of meaning (Cameron 1997: 45). Seen from the point of view of their complementary qualities and limitations the combination of the two approaches has been favourably described as a ‘natural match’ (Mautner 2009: 33) and ‘useful methodological synergy’ (Baker et al. 2008). Most literature discussing the synergy between CDA and corpus linguistics focuses on how quantitative methods can help the soundness of qualitative interpretations ² and contribute to the analysis by making it possible to handle larger amounts of texts (Orpin 2005) and by ‘checking overinterpretation and underinterpretation’ (O’Halloran & Coffin 2004). Referring to Corpus-Assisted Discourse Studies (CADS), Vessey describes it as ‘a useful framework because of the ways it addresses some of the weaknesses and exploits the strength of its constituent parts’ (Vessey 2013: 3), namely corpus linguistics and ‘traditional discourse analysis’ (Partington & Marchi 2015: 216). Corpus linguistics is ultimately seen as improving the analysis by providing more neutral starting points and generalisability, i.e. accounting for the scale of phenomena. At the same time the overall ‘qualitative’ nature of the analysis guards against the commonly lamented trouble with quantitative studies...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Contents
• List of figures
• List of tables
• Contributors
• Acknowledgements
• 1 Introduction: partiality and reflexivity
• PART A Overlooked areas (checking the dusty corners)
• PART B Triangulation (identifying blind spots)
• PART C Research design (avoiding pitfalls/re-examining the foundations)
• Index