THE CRITICAL TURN IN APPLIED LINGUISTICS
The critical turn seemed to arrive in applied linguistics in the 1990s in a call by Pennycook (Reference Pennycook1990) in his seminal article “Towards a Critical Applied Linguistics for the 1990s”:
If we are concerned about the manifold and manifest inequities of the societies and the world we live in, then I believe we must start to take up moral and political projects to change those circumstances. This requires that we cease to operate with modes of intellectual inquiry that are asocial, apolitical or ahistorical. (pp. 25–26)
The field has since witnessed increasing studies drawing on a variety of analytical frameworks and approaches, many of which come under the umbrella term of critical discourse analysis (CDA). As CDA is interested in uncovering ways in which social structures of inequality are produced in and through language and discourse, it contributes to the critical turn in applied linguistics by offering theoretical and methodological resources for critical inquiry. In the following sections, I shall first outline the theoretical assumptions and major methodological approaches of CDA. Then, the applied linguistics areas in which researchers have conducted their studies with CDA approaches are reviewed. The strengths as well as the limitations of these approaches are discussed and the latest developments of CDA methodology are outlined with an emphasis on directions for future methodological explorations that can offer the greatest potential in overcoming some of the current methodological and theoretical limitations.
CDA: THEORETICAL ASSUMPTIONS AND METHODOLOGICAL APPROACHES
CDA has a history of only about three decades, with three key researchers who have made seminal contributions: Teun van Dijk, Ruth Wodak, and Norman Fairclough. Each has used somewhat different analytical frameworks and focused on analysis of discourse in slightly different domains. CDA is thus best understood not as a unitary discipline but as a cluster of interdisciplinary approaches (e.g., van Dijk prefers the term critical discourse studies to emphasize its diversity) that are, however, united by some common theoretical assumptions and commitments. CDA's commitments are summarized in the following list (based on van Dijk, Reference van Dijk, Wodak and Meyer2009; Reisigl & Wodak, Reference Reisigl, Wodak, Wodak and Meyer2009) and later used as criteria in reviewing recent CDA studies in applied linguistics.
1. CDA is socially committed research: It is committed to contributing to the understanding and tackling of social problems, especially those that are caused or exacerbated by public text and talk, such as various forms of social power abuse (e.g., racism, sexism, classism, homophobia) and their resulting social consequences.
2. CDA is thus problem-oriented rather than theory or discipline-driven and emphasizes presenting its practical implications and applications in accessible language to the public or relevant parties.
3. CDA is interdisciplinary and calls for flexibility and diversity in its approaches and methods to tackle complex issues and problems.
4. CDA takes into account the interests, expertise, and resistance of those groups that are subjected to discursive injustice.
5. CDA stresses researcher reflexivity—the need to make the object under investigation and the analyst's own position transparent and justify theoretically why certain interpretations and readings of discursive events seem more valid than others.
Reconceptualization of Language as Social and Discourse Processes
CDA is not interested in investigating language as a static linguistic entity but in studying social and discursive processes and their consequences. It holds the theoretical assumption that language should not be seen as a reified object of study, and language treated as a bounded entity is actually an ideological and social construct, born of activities of political, nationalist, or colonial segregating agendas (Pennycook, Reference Pennycook2001, Reference Pennycook2010). The analytical focus should thus be on how language, as continuously changing systems of semiotic resources, among other semiotic systems of resources (e.g., multimodalities), are recruited and utilized for constructing racial, gender, social, sexual, and other cultural categories that legitimate and perpetuate inequalities (e.g., policies, institutions) in society. CDA maintains several assumptions about language: It sees language as social practice; language and discourseFootnote 1 both shape/constitute and are shaped/constituted by social structures (e.g., gender, sexuality, class, ethnic identities); and language is intrinsically ideological and plays a key (albeit often invisible) role in naturalizing, normalizing, and thus masking, producing, and reproducing inequalities in society.
Theoretically CDA has drawn on the social theories of Althusser (Reference Althusser1984), Gramsci (Reference Gramsci1975), Foucault (Reference Foucault, Gordon, Gordon, Marshall, Mehpham and Soper1980, Reference Fairclough2003), and Bourdieu (Reference Bourdieu1984, Reference Bourdieu1988) to analyze the working of power through language and discourse. However, CDA has extended these social theories by developing analytical tools that are informed by theories of linguistics, systemic functional linguistics, and semiosis (i.e., meaning-making processes). A brief outline of the representative approaches developed by van Dijk, Wodak, and Fairclough respectively will illustrate the multidisciplinary methodological characteristics of CDA.
Van Dijk's Sociocognitive Theory of Critical Discourse Studies
Among the key CDA researchers, van Dijk is the only researcher who attempts to integrate cognitive theories with linguistic and social theories. His sociocognitive theory of critical discourse studies (CDS) proposes a three-layer model of the interrelationships among discourse, cognition, and society (van Dijk, Reference van Dijk, Wodak and Meyer2009). Put simply, van Dijk uses cognitive schema theory (e.g., context models) as the middle layer to mediate between structures of language and discourse (e.g., lexical choices, rhetorical strategies) at the micro level and structures of society at the macro level (e.g., gender, ethnic, sexual categories). Van Dijk, however, reminds us that the micro–macro distinction is only an analytical metaphor. CDS is at the same time cognitive, social, and political analysis, and “it is this permanent bottom-up and top-down linkage of discourse and interaction with societal structures that form one of the most typical characteristics of CDS” (van Dijk, Reference van Dijk, Wodak and Meyer2009, p. 83).
Bourdieu's influence can be seen in van Dijk's analytic focus on how social inequalities and ideologies in the larger society are being reproduced and perpetuated in everyday local language and discourse practices. For example, van Dijk applies both linguistic and social psychological approaches to the critical analysis of racist discourses of the white ruling elites in European, British, Australian, New Zealand, North American, Latin American, and South African societies (van Dijk, Reference van Dijk1993). Van Dijk's study (Reference van Dijk1993) critically analyzes racist discourses from all key domains: political, corporate, academic, educational, and media. Van Dijk differentiated between elite racism and popular racism and argues that it is the racist discourses of the elites in different domains of society that provide both the cognitive frameworks (schemata) and the discursive resources for the reproduction of ethnic stereotypes in everyday talk and thought of the masses. Drawing on Bourdieu's theories (Reference Bourdieu1984, Reference Bourdieu1988) about how a society's dominating elites use their symbolic capital to legitimate and authorize their ways of speaking, thinking, writing, and acting (i.e., their habitus), van Dijk further sees these elites as playing an important role in the authorization and legitimation of racist policies and everyday racist practices (van Dijk, Reference van Dijk1993).
The strength of van Dijk's approach lies in his integration of cognitive schemata theories with discourse analysis. His provision of the middle analytical layer (e.g., a social actor's cognitive model of the context) serves to explain how larger societal structures (e.g., racist ideologies, racist policies) come to be enacted and reproduced by local social actors in everyday language and discourse practices. The weakness of his approach is the focus on analysis of reproduction rather than transformation. However, this weakness is not unique to van Dijk as this has been a critique of Bourdieu or CDA in general (e.g., Canagarajah, Reference Canagarajah1993; Jenkins, Reference Jenkins1992). In fact, van Dijk's provision of the middle analytical layer can lend his methodology to the analysis of how transformation can take place. Also, Bourdieu's later theories speak to how transformation can take place (Block, Reference Block2013), but more future CDA research needs to be devoted to studying the processes of transformation rather than just reproduction of ideologies and also to developing the methodological tools to analyze transformation processes. Wodak's prospective critique seems to be a good beginning in this direction.
Wodak's Discourse-Historical Approach
Like van Dijk, Wodak's discourse-historical approach (DHA) has been influenced by European critical social theories (e.g., Frankfurt School's critical theoryFootnote 2) and focuses on critiquing the naturalization and masking of ideologies in everyday language and discourse. She has elaborated on the following three kinds of critique (summarized from Reisigl & Wodak, Reference Reisigl, Wodak, Wodak and Meyer2009):
1. Text or discourse-immanent critique aims at discovering inconsistencies, self-contradictions, paradoxes, and dilemmas in the text-internal or discourse-internal structures.
2. Sociodiagnostic critique aims at demystifying the manifest or latent persuasive or potentially manipulative character of discursive practices. Here, the analysts draw on their contextual knowledge and social theories as well as other rhetorical models from various disciplines to interpret the discursive events.
3. Future-related prospective critique seeks to contribute to the improvement of communication (e.g., by elaborating guidelines against sexist language use or by reducing possible language barriers in hospitals, schools, etc.).
Wodak and her colleagues have developed elaborate frameworks and procedures of analysis that include the systematic collection of samples of texts (both spoken/written, and visual) on a macro topic (e.g., climate change) over a period of time. Analysis is conducted not only on texts but also on the interrelationships of discourse topics and genres found in the different political fields of actions (e.g., lawmaking procedure, formation of public opinions) related to the macro topic. Interdiscursive and intertextual relationships between Discourses, Discourse topics, genres, and texts over time are carefully charted out. This focus on how topically related networks of texts and genres evolve around a set of ideologies (i.e., Discourses, in Gee's terminology), which draw on and reinforce one another over time and across multiple texts and sites, is a key strength of Wodak's methodology. This focus of Wodak on intertextuality and interdiscursivity provides a powerful methodology to explain why macrostructures of inequality are persistent and pervasive and can get reinforced and perpetuated via discursive processes across multiple sites and multiple texts over a sustained period of time.
In text analysis, DHA, in general, follows a three-step analytical procedure. After (a) identifying the topics and texts of a specific Discourse, (b) discursive strategies are investigated. Then linguistic means (as types) and the specific, context-dependent linguistic realizations (as tokens) are examined. The following five heuristic questions usually guide DHA analysis:
1. How are persons, objects, phenomena/events, processes, and actions named and referred to linguistically?
2. What characteristics, qualities, and features are attributed to social actors, objects, phenomena/events, and processes?
3. What arguments are employed in the discourse in question?
4. From what perspectives are these nominations, attributions, and arguments expressed?
5. Are the respective utterances articulated overtly; are they intensified or mitigated?
Wodak's methodology is concrete and operationalizable (i.e., can be realized as a set of concrete analytical procedures). This is perhaps both a strength and a weakness. The strength is that the analytical procedure is easy to follow by someone with a linguistic background but not a social theoretical background. The weakness perhaps lies in the largely linguistic analytical focus that does not also equip the analyst with the social theoretical frameworks to connect the linguistic analysis to the analysis of social practice (e.g., in the fields of action) as well as larger social structures. This seems to be a project that Fairclough has embarked on.
Fairclough's Dialectical-Relational Approach
Fairclough's dialectical-relational approach (DRA; 2009) aims at integrating social theory, practice theory, and linguistics theories. As the term discourse is commonly used in different senses ranging from (a) meaning-making as an element of the social process, (b) the language associated with a particular social field or practice (e.g., political discourse), and (c) a particular social perspective or ideology (e.g., neoliberal discourse of globalization), Fairclough prefers to use the term semiosis for the first, most abstract sense. Fairclough draws on both systemic functional linguistics (SFL) theories of semiotics (i.e., meaning-making systems), conversation analysis, and other pragmatics theories in developing text analysis methods; for example, analysis of transitivity, modality, turn-taking, exchange structure, formulation, politeness, and topic control (Fairclough, Reference Fairclough1992), and analysis of discourses, genres, and styles/voices (Fairclough, Reference Fairclough1995). His early work analyzed medical interview discourses as well as political discourses and news media discourses. The most recent work of Fairclough and his colleagues drew on argumentation theory to develop practical frameworks for analyzing argumentative texts, in particular, for uncovering the neoliberalist ideologies embedded and masked in the government discourses and economic crisis discourses (e.g., in government budget reports, parliamentary debates) circulating in the public spheres in the United Kingdom (Fairclough & Fairclough, Reference Fairclough and Fairclough2012).
Fairclough has developed a three-layer model of social reality: social structures, practices, and events. Analysis focuses on two kinds of dialectical relations: (a) those between social structures and events (emphasizing the mediating role of practices between the two), and (b) within each event, those between semiotic and other elements. Semiosis is seen as an element of the social process, which is dialectically related to other elements. Relations between elements are dialectical in the sense that they are different but not discrete or fully separate. Each internalizes the others without being reducible to them. Semiotic analysis, in turn, makes use of three analytical categories: genre, discourse, and style. To Fairclough, CDA oscillates between a focus on structures (especially the intermediate level of the structuring of social practices) and a focus on the strategies of social actors (i.e., the ways in which they try to achieve outcomes or objectives within existing structures and practices, or to change them in particular ways; the latter will lead to social change). The strength of Fairclough's frameworks lies in its ambitious theorizing to connect microlinguistic analysis to analysis of social practice (intermediate analytical level) and analysis of social structures (macro-analytical level). This framework allows better provision for the theorizing of how transformation can take place. It has also become particularly relevant to the recent focus in applied linguistics on analysis of the global spread of neoliberalism and its incursion in education (e.g., Block, Gray, & Holborow, Reference Block, Gray and Holborow2012; Duchêne & Heller, Reference Duchêne and Heller2012; Heller, Reference Heller2011). Its weakness (or not, depending on one's perspective), perhaps, is its highly complex system of terminology and relative theoretical difficulty to someone with only an applied linguistics background but not also a social theoretical background. Researchers drawing on Fairclough's analytical frameworks need to be able to translate the findings in language accessible to the general public in order for it to have a wider social impact.
It can be seen from the preceding synoptic outlines that although all the three key CDA scholars have developed diversified approaches and frameworks, they share some common assumptions and commitments. For example, their frameworks all (albeit to varying extents of theoretical delicacy) provide for some theorizing about the relationship between microstructures of language/discourse/events and macrostructures of society, with a mediating middle layer (e.g., social practices, context models, or fields of action), and they also caution that the micro–macro divide is merely an analytical metaphor, not to be reified. For text analysis, they all draw on information about the context, albeit with somewhat different theorizing about the context. It might be fair to say that CDA researchers attempt to further the socially committed work of critical, social theorists by integrating the theoretical and analytic resources from linguistics, semiosis, and rhetorical theories. Cognitive theories and corpus linguistics tools are also drawn upon by some researchers (van Dijk, Reference van Dijk, Wodak and Meyer2009; Mautner, Reference Mautner, Wodak and Meyer2009). And a CDA study usually involves the “linguistic description of the language text, interpretation of the relationship between the (productive and interpretative) discursive processes and the text, and explanation of the relationship between the discursive processes and the social processes” (Fairclough, Reference Fairclough1995, p. 97; original emphasis). It must be pointed out that diverse linguistic theories can be drawn upon to do the linguistic description, but just doing the linguistic description will not qualify the description as CDA. The description must also be linked to the interpretation and explanation of the relationship between discursive and social processes.
AREAS IN APPLIED LINGUISTICS IN WHICH CDA METHODS HAVE BEEN APPLIED
In applied linguistics, CDA methods have been applied in analysis of public and media discourses (e.g., press discourse, political discourse, popular cultural texts) and educational discourses. The latter include a diverse range of topics: educational policies; classroom interactions; textbooks and curriculums; interviews, narratives, and creative writings and texts produced by, and discourses of and on, students, teachers, school administrators, parents and caretakers, psychologists, education policymakers, teacher-educators, community leaders, new immigrants, and so on.
In this article, recent studies in the following three main areasFootnote 3 will be reviewed: textbook/curriculum text analysis, discursive positioning analysis, and public and media discourse analysis. In selecting these studies, I searched for articles containing the key words critical discourse analysis from 2008–2013 in the following seven international journals in the field of applied linguistics: Applied Linguistics; Linguistics and Education; Language and Education; TESOL Quarterly; Language, Culture, and Curriculum; Pedagogies; and Critical Inquiry in Language Studies.
From the search results, I selected studies in light of three general considerations: (a) the studies explicitly claim to be doing CDA or using CDA methods; (b) the studies come from diverse sociocultural contexts; (c) the studies can illustrate a range of issues that are of theoretical and methodological interest.
Given limited space and the diverse range of studies that needs to be covered in this review, in what follows, I shall present summaries of the studies under three major headings and then comment on the methodology.
Textbook and Curriculum Texts Analysis
In this section I shall present summaries of three recent studies (de los Heros, Reference de los Heros2009; Gulliver, Reference Gulliver2010; Le Roux, Reference Le Roux2008) that analyze textbook and curriculum texts from three contexts: Peru, Canada, and South Africa.
The first study by de los Heros (Reference de los Heros2009) focuses on examining a single textbook: Talent: Language and Communication (Talento: Lengua y communicacion, 2004). This textbook is chosen because it is the official Peruvian language textbook for first-year high school students. De los Heros situates her critical analysis in the larger education policy context of Peru. Recent policy reforms as espoused in Education Law No. 28044 (Ministerio de Educación, 2005) have established the teaching of respect for indigenous languages and language diversity as a main goal in the area of language education in Peru. However, de los Heros finds that the textbook embeds language ideologies that run against this policy. Specifically, she finds that the textbook evidences a hidden curriculum that advances linguistic prescriptivism and the superiority of standard Spanish over regional varieties of Spanish. The author mentions the use of Fairclough's CDA frameworks (Reference Fairclough1995, Reference Fairclough2003) and SFL. Analysis of the stigmatization of regional varieties of Spanish and their speakers is presented through analysis of specific lexico-grammatical choices. For example, barbarism is used to refer to stigmatized forms, or deformed language to characterize Spanglish. The first-person plural verb endings are also frequently employed to create the impression that students are themselves involved in the learning and decision process. Also, rhetorical questions and mini-dialogues are used to simulate an exchange of ideas between the author/writer and the reader/student when that is not really the case as the text is actually very prescriptive and authoritative (de los Heros, Reference de los Heros2009).
This kind of feigned dialogicality is also found in the analysis by Gulliver (Reference Gulliver2010) of immigrant success stories in 24 English as a second language (ESL) textbooks that are used in the government-funded immigrant education programs in Canada. Gulliver identified two primary strategies of legitimation: positive self-representation/negative other-representation and the maintenance of a low orientation to difference (Fairclough, Reference Fairclough2003), despite a feigning of dialogicality through highly inauthentic constructed dialogues. These stories all follow a very similar narrative pattern that legitimates periods of economic hardship, social exclusion, and personal struggles, but overwhelmingly represents hard-working immigrant newcomers as successful and appreciative of the opportunities provided by Canada. Through repeating this storyline in the textbook stories, Canada's role as a redeemer of immigrants is repeatedly asserted.
Le Roux's study (Reference Le Roux2008) of one real-world math word problem in a foundation-year university mathematics course in South Africa aiming to provide marginalized students with access to tertiary studies in the sciences also mentions the use of Fairclough's CDA framework. In recent math education discourses in South Africa, access and relevance are two notions being emphasized and through the use of real-world word problems, it is supposed that math and science knowledge will be made more accessible and relevant to a greater range of students. However, le Roux found in the real-world math problem that she analyzed, traces of texts from six different Discourses. She argued that instead of making math learning easier, the real-world word problem actually requires the student to have implicit knowledge of the school mathematical word problem genre and the associated assumptions. It also requires students to buy in some of the capitalist Discourse assumptions embedded in the real-world word problem.
All of the preceding three studies focus on analysis of textbook texts, and sometimes only one single text is analyzed. This single-text approach is good at illustrating an analytical method, for example, le Roux's (Reference Le Roux2008) method of identifying the traces of other Discourses in the real-world problem text and de los Heros's (Reference de los Heros2009) use of SFL-inspired linguistic analysis to analyze a textbook. However, in terms of furthering the critical project of CDA, single-texts approaches are useful mainly as a proto-analysis or a pilot study and need to be followed up by subsequent studies that systematically sample a broader range of the curriculum materials in question and over time (c.f., Wodak's DHA; see Reisigl & Wodak, Reference Reisigl, Wodak, Wodak and Meyer2009). To review these studies, let us revisit the five commitments of CDA listed at the beginning of this article:
1. CDA is socially committed research.
2. CDA emphasizes presenting its practical implications and applications in accessible language to the public or relevant parties.
3. CDA is interdisciplinary and calls for flexibility and diversity in its approaches and methods to tackle complex issues and problems.
4. CDA takes into account the interests, expertise, and resistance of those groups that are subjected to discursive injustice.
5. CDA stresses researcher reflexivity.
It seems that the preceding textbook studies are all socially committed and can present their findings in language accessible to the public. However, they seem to have focused mainly on the naturalization and masking of ideologies in textbook texts without also collecting data on how these textbook texts are actually used and taken up by the school participants in different ways (e.g., any evidence speaking to their possible resistance to, or alternative interpretations of, these texts). This leads to the methodological comment on the limitation of a singular focus on textual analysis without integrating textual analysis with a more ecological approach (e.g., Kramsch, Reference Kramsch2002; van Lier, Reference van Lier2004; Rogers, Reference Rogers2012). There is also the need for more researcher reflexivity in these studies (e.g., the need for more discussion of the analyst's own sociopolitical position and why certain interpretations and readings of the text seem more valid than others; why some linguistic analytical tools, such as those from SFL, are used; what the criteria are for the selection of some tools and exclusion of some others in the study). However, to the credit of these studies, they have all shown that the main contribution of a detailed textual analysis lies in the insight that can be gained in the detailed linguistic patterns showing how language processes are central to the operation of hidden curriculums, which have previously been analyzed mainly from sociopolitical perspectives (Apple, Reference Apple1979).
Discursive Positioning Analysis
In this section, I shall present summaries of two recent studies that focus on discursive construction of identities and discursive positioning. They are selected to discuss issues of theoretical and methodological interest.
The study by Briscoe and de Oliver (Reference Briscoe and De Oliver2012) of the interview discourse of 22 public school leaders (half coming from the three highest-wealth school districts and half from the three lowest-wealth school districts) in one of the largest cities in Texas is situated in the context of the No Child Left Behind (NCLB) policy, under which financial resources are allocated to public schools according to students’ standardized test scores. The authors first present the 2011 U.S. Census data that show the disproportionately high representation of Hispanic and African American groups in the low-income and low-education brackets. The authors invoke Foucault's (Reference Foucault, Gordon, Gordon, Marshall, Mehpham and Soper1980) notions of macrostructures and practices to argue that the sorting mechanisms of schools (which are examples of macrostructures and macro practices) emerge from micro practices, including discourse. They thus set out to analyze Texas school leaders’ discursive construction of minority and low-income family identities within the neoliberal context of NCLB, aimed at discovering whether the family identities that school leaders produced and reproduced through discourse supported or countered ideologies of racism and/or classism.
Using qualitative content analysis, the interview texts are coded by two researchers for themes related to families and from them, the researchers identify a dominant Discourse: Families are responsible for their children's success or failure in schools. The authors, however, have not provided a discussion of the interpretive processes of the coders in coding the interview text. More researcher reflexivity is needed, especially in problematizing the role of the interviewer in the research interviewing process in partially shaping/constituting the data that is produced and subsequently interpreted and coded by the researchers, as the interviewing process is in itself a social practice (Mishler, Reference Mishler1991; Talmy, Reference Talmy2010, Reference Talmy2011).
The researchers also checked the academic journals to see if they had any themes in common with the dominant Discourse of the school leaders and they found an article in a leading academic journal that supported both the dominant Discourse and each of the five interrelated and supporting derivational themes identified in the content analysis of the interview data. The themes of the dominant Discourse were found to be aligned with neoliberalist ideological presumptions of choice, individualism, meritocracy, and a level playing field.
The researchers, however, also identified a counter Discourse in some of the interview data that contested and resisted the construction of deficit identities for traditionally underprivileged families. The authors concluded that the dominant Discourse that Families are primarily responsible for their children's success or failure in school seems innocuous and innocent, harkening to neoliberalist ideologies of rugged individualism. These deficit constructions of low-income and minority families are not merely compatible with, but actually supportive of, social practices that lead to the current stratification of schools and society. However, the authors also identified a counter discourse in some school leaders’ interview texts. Thus, the authors provide the reminder that individuals and the discourse they produce are not totally determined by power/relations, stressing the agency of some school leaders in countering the larger social structures of racism and classism.
Using the five commitments of CDA to review this study, we can see that this study is socially committed research and it considers the agency of the participants and presents its findings in accessible language. However, it falls short of the criteria of researcher reflexivity and the use of a range of research methods other than just content analysis in its analysis of the interview process and data.
The second study included in this section is Gu's case study (Reference Gu2010) on the respective interactions of two Chinese female students with their English conversation partners in regular biweekly dyad sessions set up for the purpose of improving their English speaking proficiency. The two students (Jocelyn and Lisa—pseudonyms) were studying in a leading university in China while the English conversation partners (Katherine and Karin—pseudonyms) came from an international school in the city where the university is located. The two conversation partners both happened to be Christians and had come from the United States. All four participants were recruited on a voluntary basis for this research project set up by the researcher. The conversation program lasted two academic semesters. Interview and diary data were also collected from the two Chinese students (but not from their conversation partners) to compare and contrast with the audiotaped dyad interaction data. The study asked the following research question: “In what ways do the sociocultural discourses in mainland China and the learners’ histories as well as their imagination influence their identity construction in a dyadic interactional context?” (p. 302)
Gu mentioned using the analytical tools from Fairclough (Reference Fairclough2003) to examine the construction of identities through deontic modality, epistemic modality, and evaluation. Deontic and epistemic modality are linguistic features that indicate speakers’ certainty with and commitment to necessity, obligation, truth, and the reliability of the sources of knowledge in one's claims (Fairclough, Reference Fairclough2003). In addition, Gu also analyzed learners’ discursive strategies in constructing identities, and referenced the “logic of equivalence” (Laclau & Mouffe, Reference Laclau and Mouffe2001, as cited in Fairclough, Reference Fairclough2003, p. 87), which is established by placing words of similar ideological import throughout one's text to strengthen a particular ideological effect (Fairclough, Reference Fairclough2003, p. 310).
Gu concluded that the identities of Jocelyn and Lisa are constructed through their discursive strategies of establishing oppositions and differences. Jocelyn's identity as a Christian gives her legitimacy in her interaction with Katherine, but through her constantly comparing and constructing two cultures, binary oppositions are constructed. In her discursively constructed oppositions, Jocelyn constructed conflicting identities, such as a Chinese person who loves her mother culture, a simultaneous interpreter, and so on, as she provided changing evaluations about, and developed different degrees of affinity with the cultures in these two broad categories.
While Gu's study provides convincing analysis of the discursive ways in which the two Chinese students negotiate their identities during their sometimes not entirely comfortable encounters with their American conversation partners, Gu's application of CDA methods has, somehow, not enabled Gu to detect Jocelyn's discursive construction of homosexuality as a social problem, along with other stereotypical representations of the social problems in the United States and China. In the analysis of Jocelyn's gradual process of reaching a relatively balanced view of self and other (i.e., Chinese and U.S. societies both have strengths and weaknesses), Gu (Reference Gu2010) wrote:
Jocelyn gradually developed a new view of the Chinese and American education systems and societies, without particularly favoring either one. She said, “I think that China and American societies both have problems. In China, the moral education and democracy are not good, but in the United States, the social problems such as the teenager crimes, drugs and gays or lesbians are very serious. . .” (Interview 3, month 12). Jocelyn's commitment to the statement was expressed through the use of the present tense, ‘‘I think that China and America both have problems.’’ In the second sentence, Jocelyn used the categorical language are when pointing out the disadvantages of the Chinese and American education systems, without using hedges such as I think or I feel, so expressing a high degree of affinity with truth of the claim. . . . . In the text, Jocelyn also employed evaluative words like not good, serious, very strong, and very weak to express her degree of commitment concerning what is good or bad. . . . . Instead of choosing one culture as superior, Jocelyn compared them and pointed out their differences. (pp. 318–319)
Gu's preceding analysis shows that her exclusive focus on the linguistic features of deontic modality, epistemic modality, and evaluation in the utterances seemed to have distracted her from noticing other ideological aspects of the utterances and the dominant Discourses circulating in both the official and public media in China that Jocelyn seemed to be drawing on in order to build her balanced view of self and other (more on this later). An adherence to a specific linguistic analytical method can produce blind spots in the analyst. A heightened sensitivity to the ideological aspects of text is, however, difficult to achieve as text interpretations are filtered through the subjectivity of the researcher (Fairclough, Reference Fairclough2003), and in this case, the researcher herself is likely to have been immersed in and subjected to the ideological effects of dominant Discourses circulating in the Chinese public media. Being reflexive is thus easier said than done and needs to be taken seriously by critical discourse analysts (Reisigl & Wodak, Reference Reisigl, Wodak, Wodak and Meyer2009). Using the five CDA commitments to review this study, it can be said that while the study aims to be socially committed and analyzed the agency of the research participants in constructing more favorable learner identities for themselves, it falls short of the criteria of researcher reflexivity and the use of interdisciplinary research methods and perspectives; for example, not just focusing on certain linguistic features but also paying attention to the larger problematic Discourses that are being reproduced in the participants’ discursive constructions of binary, essentialized identities of different social and cultural groups. While some isolated aspects of Fairclough's CDA frameworks are cited, the central commitments of CDA have not been fully or firmly grasped.
Public and Media Discourse Analysis
In this section I shall present summaries of two studies, O’Halloran (Reference O’Halloran2007) and Baker, Gabrielatos, and McEnery (Reference Baker, Gabrielatos and McEnery2013). O’Halloran (Reference O’Halloran2007)Footnote 4 provides a methodological delineation of how corpus-based tools can help CDA researchers reduce overinterpretation of linguistic data as metaphorical. The case in point is Lee's (Reference Lee1992) claim about the use of metaphors of natural forces to represent Soweto protesters in a hard news article that appeared in the British newspaper, The Guardian, in 1976, with the ideological effect of taking away the Soweto marchers’ agency through metaphorically implying that they are like rivers sweeping through or volcanoes erupting (e.g., using words like erupted, swept through). Using evidence from a 260-million-word corpus of British newspapers, O’Halloran showed that readers of British newspapers are likely to have been exposed to natural forces metaphors in the register of hard news, and words like erupt, erupts, erupted have co-occurred with human agents as their subjects in the hard news register. O’Halloran (Reference O’Halloran2007) commented on the overall methodology of Lee (Reference Lee1992) as follows:
Looking overall at Lee's interpretation of simmering, erupted, and swept through, there is inconsistency in the way he identifies the linguistic data as metaphorical. The corpus evidence shows that Lee has a lexicalized interpretation of both simmering and erupted but has a delexicalized reading of swept through (in not relating the data to a broom, etc.). So it is not only Lee's analysis and interpretation of individual bits of data that is problematic but also how he makes overall interpretation of the data. (p. 19)
As interpretation is a key analytical process in CDA, CDA researchers’ interpretive procedures need to be given much more systematic and critical attention. Interpretive disciplines (e.g., interactive sociolinguistics, linguistic anthropology, text analysis) belong to what Habermas (Reference Habermas1987) called reconstructive sciences, whereby the largely invisible, everyday, mundane, semiotic (meaning-making) resources and procedures of members of a community are systematically reconstructed to make them explicit and visible and thus open to examination, critique, and change (or intervention). Traditionally, interpretive researchers rely on their own interpretive resources (via claims of membership in the communities in question) to derive text interpretations as well as the possible and plausible interpretations of the readers (or conversation participants) based on textual evidence (and contextual evidence). As CDA draws heavily on the interpretive disciplines for its analytical methods, corpus-based tools can certainly help to strengthen the systematicity of the interpretive reconstructive procedures by providing evidence on the range of semiotic resources specific to different registers and genres necessary for both text production and interpretation in certain communities (e.g., the meaning potential of specific words and collocates in specific registers such as the hard news register in British newspapers).
In a similar vein, Baker et al.'s study (Reference Baker, Gabrielatos and McEnery2013) used methods from corpus linguistics and CDA to examine patterns of representation around the word Muslim in a 143-million-word corpus of British newspaper articles published between 1998 and 2009. Analyzing the noun collocates of Muslim, the researchers found the following categories (in order of frequency): ethnic/national identity, characterizing/differentiating attributes, conflict, culture, religion, and group/organizations. The conflict category was found to be particularly lexically rich, containing many word types. Following this, an analysis of the two most frequent collocate pairs: Muslim world and Muslim community showed that they were used to collectivize Muslims, both emphasizing their sameness to each other and their difference to The West. Muslims were also represented as easily offended, alienated, and in conflict with non-Muslims.
What is methodologically interesting about this study is that corpus tools are used to test an initial hypothesis, and then qualitative analysis is used to generate further hypotheses, which are then tested in further corpus-based analysis. This flexible intertwining use of methods are driven by the researchers’ evolving understanding of the phenomenon and systematic generation of further hypotheses and counter-hypotheses, and subjecting them to further testing by employing both corpus analysis of specific lexical items/collocates and qualitative analysis of longer excerpts of texts.
Using the five CDA commitments to review the preceding two studies, it can be said that both studies are socially committed; both are problem-oriented and use accessible language to present their findings; both pay attention to issues of researcher reflexivity and use a diverse range of methods including both qualitative, quantitative, and corpus methods, although both studies also need to include a consideration of the agency of the readers of these media texts. In this respect, Fairclough and Fairclough (Reference Fairclough and Fairclough2012) provide good examples of how political and media discourses can be analyzed with a consideration of the structure—agency dialectic and propose “practical reasoning as an interface between agency and structure” (p. 244). While couched in different terms and drawing on different traditions, their proposal is actually akin to van Dijk's proposal of the individual's contextual models (cognitive schemata) as an interface between structure and agency. Also, CDA studies in applied linguistics would benefit from drawing more on Wodak's DHA, especially her emphasis on collecting data over a period of time, highlighting the importance of the longitudinal, historical dimension of CDA.
FUTURE DIRECTIONS AND ISSUES IN METHODOLOGICAL DEVELOPMENT OF CDA IN APPLIED LINGUISTICS
There are a number of future directions for CDA in applied linguistics. As mentioned earlier, future CDA methodological development can benefit from drawing on corpus-based tools. For example, Cheng (Reference Cheng and Chapelle2013) provides a review of corpus-based/informed CDA studies in the last two decades, noting that many of the corpora used in CDA studies so far are primarily those of media and political discourses. Cheng proposes that future studies could explore corpora of a wider range of text types (e.g. feedback commentaries in education, academic classroom discourse, business discourse). This leads us to the recent studies of translanguaging and translingual practice (Canagarajah, Reference Canagarajah and Li2011, Reference Canagarajah2013a, Reference Canagarajah2013b; Creese & Blackledge, Reference Creese and Blackledge2010; García, Reference García2005, Reference García2009; Li, Reference Li2011; Lin, Reference Lin, Blackledge and Creese2014) and the sociolinguistics of globalization; for example, the study of multilingual, multimodal, hybrid, super-diverse, super-vernacular texts, especially those found in the new media (Blommaert, Reference Blommaert2010; Blommaert, Collins, & Slembrouck, Reference Blommaert, Collins and Slembrouck2005; Blommaert & Varis, Reference Blommaert and Varis2011). The rise of globalized new media also points to the increasing need for multimodality analytical tools. Van Leeuwen (Reference van Leeuwen2008) points out that global media allows content to be diverse and localized, but homogenize formats and genres to an unprecedented degree. Van Leeuwen's (Reference van Leeuwen2008) study of how a social practice (e.g., a child starting to attend school) can be recontextualized and represented in different genres and formats (e.g., a newspaper article, a manual written for mothers), privileging different sets of assumptions about the world (ideologies), illustrates well how methods of multimodality analysis can be fruitfully applied in CDA, and the sociolinguistics studies of globalization can fruitfully interilluminate CDA studies.
We can also draw on the methodological insights of recent research on discourse itineraries (Chun, Reference Chun2010; Jones, Reference Jones, Bhatia, Flowerdew and Jones2008; Scollon, Reference Scollon, Bhatia, Flowerdew and Jones2008), which focus on longitudinal analysis of the dynamic, evolving trajectories of mediated discourses, actions, events, objects, and bodies (human bodies) in explaining the working of ideologies and power through these trajectories (called discourse itineraries by Scollon). In the same vein, as practice is a central notion in CDA that mediates between micro and macro levels of analysis, timescales analysis, which seeks to give a more refined description of practice is needed:
This strategy for moving beyond macro and micro can only succeed, however, if “practice” can be clearly defined. One sometimes hears the word invoked as if we knew the fundamental level at which social life operates—even though “practice” is used to describe widely differing timescales. (Wortham, Reference Wortham2012, p. 132).
For example, Agha (Reference Agha2007) described how the association between the phonological regularities now described as Received Pronunciation (RP) and a speaker of Standard English did not exist at the beginning of the 19th century, but was well established across much of England a century later. Agha argued that such change occurs through speech chains, a linked series of speech events in which the hearer in one event goes on to circulate a particular sign or model when speaking in a subsequent event. If more of the CDA studies in applied linguistics can move beyond the single-text, single-shot approaches (as in the example of Foucault's analyses of the genealogies of specific discourses, knowledges, practices, and social structures; see Foucault, Reference Foucault, Rabinow and Rose2003), much more interesting findings about the workings of ideologies in the field of applied linguistics can be revealed and denaturalized. The notions of timescales, discourse chains, and discourse itineraries will further enhance the methodological repertoire of Wodak's DHA, which seeks to analyze discourse over time. It will also enhance and enable the critical discourse analyst's understanding and tracing of how micro practices (e.g., a linked series of speech events) repeated (and circulated across multiple sites and contexts) over different scales of time can sediment into macrostructures of inequality (e.g., the institutionalization of RP).
Finally, in many recent CDA studies in applied linguistics, some important dimensions of subjectivity such as fantasy, pleasure, repression, fear, and desire are seldom addressed. How discursive practices, social practices, and societal structures mutually embed and interact to construct and instill certain kinds of language desire (e.g., desire for standard language, or RP), for example, should be important research questions in CDA studies in applied linguistics. Recent work by Kulick (Reference Kulick, Holmes and Meyerhoff2003), Kramsch (Reference Kramsch2009), Motha and Lin (Reference Motha and Lin2014), and Takahashi (Reference Takahashi2013) provide theorizations and delineations of the notions of subjectivity, affect, fear, and desire, which can be further drawn upon in future CDA studies in applied linguistics.
As CDA is gaining currency in applied linguistics, an increasing rhetoric of CDA marks some of the recent studies in applied linguistics. This perhaps warrants a critical reflexive CDA project on the rhetoric of CDA itself. If academic research is about knowledge production and its legitimation in the larger context of academic institutions in society, then CDA is in urgent need of some up-front critical reflexive studies and action on its own status as an academic discipline so that CDA does not lose its critical edge and avoids being domesticated in the discourse/practice/social structure matrix amid the increasingly neoliberalized academic ethos, policies, and institutions spreading worldwideFootnote 5 (Apple, Reference Apple2001; Block et al., Reference Block, Gray and Holborow2012; Heller, Reference Heller2011; Luke, Reference Luke2008).
ACKNOWLEDGMENT
The author wants to thank the editor and an anonymous reviewer for their very useful, critical comments on an earlier draft of this article.
