Critical discourse analysis in new media: theoretical and methodological challenges

This paper investigates the range of theoretical and methodological challenges facing scholars attempting to do studies employing Critical Discourse Analysis (CDA) in the context of New Media, such as participatory web platforms, social networking websites, or online forums. The rise and popularity of Web 2.0 has attracted researchers from diverse fields of academic inquiry to this area. However, CDA scholars had not paid attention to these domains until recently. The specific features of Web 2.0 spaces create a multitude of challenges for a (critical) discourse analyst, ranging from issues of language, to problems of data collection and applicable theoretical frameworks. Drawing from the existing literature, and also experiences gained through a CDA study on Facebook discourses, this article will discuss these challenges, the current state of affairs, and the limitations of doing discourse analytical studies in New Media. Keyword: Critical discourse analysis; New media; Web 2.0; Social media; Participatory web platforms


Introduction
The role of the internet in today's life cannot be denied.Great changes have been made with the introduction of the World Wide Web into human life.The speed of communication, the easier spread and flow of information, the ease and speed of getting things done, are all owed to advances in the field of online and electronic communication.A second and perhaps larger leap towards more changes appeared with the emergence of what is known as Web 2.0.Unlike the first generation of websites, Web 2.0 allowed a more interactive, user-centred experience of the internet, in which users could easily create content and share it with the world around them (Greaves & Mika, 2008;Murugesan, 2007).
In this new era, an internet operator was not merely a 'user' anymore.Each individual can now 'produce' content on the web, and at the same time, use the content produced by other individuals on the internet.This new culture of 'produsage' has affected many, if not all, aspects of human life today (Bruns, 2007).
Naturally, due to these changes, members of academia have also paid due attention to these domains.Scholars from various disciplines have taken up research projects to make things run even smoother, to understand the dynamics of Web 2.0, and investigate the effects it has had on social aspects of human life.Although these endeavours root from an array of academic principles, hence drawing from distant epistemologies and perspectives, they all come together in the issue of data.
Within social sciences, linguists have also been attracted by the research opportunities of Web 2.0.However, because of the potential differences between what can be named 'real' communication, taking place in the civic and communal life and in a face-to-face fashion, and the 'virtual' communication, which primarily happens in online spaces, these researchers have felt the need for newer methodologies, theoretical and conceptual frameworks, tools and methods, and modes of interpretation and analysis.Androutsopoulos's Discourse-Centred Online Ethnography (Androutsopoulos, 2008), or Herring's Computer-Mediated Discourse Analysis (Herring, 2004) can be mentioned as examples of these new methodologies.
Although linguists and other social scientists have been interested in Web 2.0 spaces for a relatively long time, it appears that discourse analysts, and especially critical discourse analysts, have shown a somehow late enthusiasm towards the New Media (Mautner, 2005).This means that the range of devices available in critical discourse analysis at present cannot be deemed satisfactory for a research in these spaces.Fortunately, more and more researchers can now be seen who are interested in, and trying to, develop the necessary tools, methods, methodologies, theories, and frameworks to be applied by CDA researchers in Web 2.0 studies.Nevertheless, there is still a lot to be done, and there are still challenges to be addressed.The present paper is the product of the experiences gained in a CDA study in the New Media, as well as a compilation of the literature at hand on the issue.Our goal is to provide future researchers aiming to work in these domains with a clearer picture and understanding of the limitations of doing CDA studies in the New Media, the challenges they are likely to meet, and the requirements they need to satisfy.We will point to two major considerations in this regard, one being the issues concerning theories and theoretical frameworks, and the other being some methodological challenges and limitations.

CDA: Approaches and Points of Interest
Critical Discourse Analysis has traditionally been a 'leftist' approach to the study of language and discourse.That is, the major theories and theoretical frameworks influencing CDA scholars all originate from, or have clear links with, Marxist traditions (Fairclough, 1989(Fairclough, , 1995)).The various approaches and methodologies developed by different scholars explicitly point to the bigger theoretical frameworks they work under.For instance, Norman Fairclough's socio-dialectical approach to CDA (Fairclough, 1989) employs many of the Marxist notions in its analyses, including Marx's theories of the ruling class and the middle class, dominant ideologies, and Althusser's theories of state apparatus (Althusser, 2006), among others.In a similar vein, Ruth Wodak's discourse-historical approach is theoretically in line with the Critical Theory, especially the Frankfurt school, which is itself a Marxist tradition in cultural theory (Reisigl & Wodak, 2001, 2009).Given this fascination with leftist theories, it is naturally expected that the usual points of interest for CDA scholars and researchers are areas of social experiences directly dealing with struggles, social problems, ruling ideologies, etc., making CDA a problem-oriented research discipline (Wodak & Meyer, 2009).Therefore, issues such as discrimination, sexism, racism, asylum seekers, refugees, xenophobia, and alike are the usual sites of CDA research, and a huge bulk of studies done from a CDA perspective are carried out on similar issues in various settings and contexts.
Because of this very leftist nature, and also due to the pioneers' interest in such issues, a majority of other researchers have taken the same perspectives, studying and analysing various discourses and texts, from political speeches to medical practices.However, although the range of topics of such studies might suggest diversity, a deeper review of literature reveals that a majority of CDA studies have only focused on the discourses of the 'elite'.In other words, most studies done from a CDA perspective have analysed the discourse of the ruling ideology, such as presidents, managers, teachers, doctors, policy makers, parliament members, etc. (Khosravinik, 2014).In short, most CDA studies have taken a top-bottom approach, in the sense that the discourse of the 'top', or the 'powerful', has been given much more attention than the discourse of the ordinary people, the dominated groups, or in Marxist terms, the middle and the working classes, or the proletariat.
The aforementioned gap in the literature, or the dearth of studies employing a bottom-up approach to ideologies and relations of power, might partly be because of the issue of data accessibility.In any given discriminatory context, the ruling class, the dominant ideology, or the elite have an advantage in creating 'content ' (van Dijk, 1988).That is, through controlling the more traditional sorts of media, such as the radio, the press, and the television, the dominant ideology is free to disseminate information and its ideologies in a large scale (Adorno & Horkheimer, 2001;van Dijk, 1993).For a researcher interested in CDA, these discourses are highly accessible and easily collectible.On the other hand, accessing discourses of the dominated groups, who do not have any tribunal to voice their opinions, requires much more resources, such as manpower, time, and finances.Additionally, the so-called 'gatekeepers' in various spheres are another problem in the path of collecting such data, since a common practice of the dominant ideologies is controlling the levels of 'access' (Fairclough, 1989;van Dijk, 1999).In other words, researchers will have to address a greater number of obstacles on their path to collect the discourses of the dominated groups.
Nevertheless, the rise of Web 2.0, and following that, the increasing popularity and spread of participatory web platforms or social networking sites, such as the Facebook or Twitter, has had a huge influence on the issue of access (Bruns & Burgess, 2011;Murugesan, 2007).At present, any individual with a smartphone and an internet connection has a voice (Unger, 2012).Therefore, the issue of accessing the discourses of the dominated groups has been made much easier.For this very reason, more CDA studies with a bottom-up approach can now be found.However, although it is true that accessing data is much easier through Web 2.0, the point that should not be overlooked is that this rise has also created a drastic challenge for the existing theories and methodologies (Kelsey & Bennett, 2014;Khosravinik, 2014).In the following sections, we will go through these challenges and limitations, by first considering the theoretical issues to be addressed when doing a CDA study on these new discourses, and then moving to some problems of methodologies and analytical tools.

CDA and New Media: Theoretical issues
The New Media, including weblogs, Social Networking Sites (SNSs), and generally, participatory web platforms, have unique features which have influenced not only the communication patterns of people, but also the dynamics of social practices, flow of information, power relations, and content creation (Dahlgren, 2005(Dahlgren, , 2009)).Previous theorizations regarding such issues lack satisfactory tools and explanations for these new dynamics.The three main features distinguishing New Media from other mainstream media are the different patterns of power relations (Khosravinik, 2014), information flow (Katz & Lazarsfeld, 1956), and time relations (Unger, 2012).CDA researchers working in New Media spheres should consider these features and peculiarities of the New Media.We will look into each of them below.

Power Relations and the New Media
Historically, power was perceived to be something possessed by the sovereign, the ruling class, or the so-called elite in a given society (Tilly, 1991).Until mid-20 th century, almost all theorizations of power and power relations viewed it as something owned by the 'top' sections of a society, flowing down to the general public, and forming their worldviews, ideologies, and practices.In such an understanding of power relations, mainstream media was always seen as a strong and effective tool in the hands of the elite, through which they could reproduce their ideologies, manufacture consent, and exert power (Adorno & Horkheimer, 2001;Althusser, 2006).Parallel to these ideological tools, the repressive tools employed by the sovereign to exert power and silence the opposing ideologies were seen to be forces such as the army, the police, and the prison system (Althusser, 2006;Tilly, 1991).Newer theorizations of power relations emerged through 1960s and 70s, with theories such as Foucault's conceptualizations of discourse and power, viewing power relations not only within the media and the repressive forces of the state, but also flowing in all circles and spheres of society, such as medicine (Foucault, 1963), policies of population control (Foucault, 1978), discourse of mental disorders (Foucault, 1963(Foucault, , 1977)), and even sexuality (Foucault, 1978).In these new understandings, power was theorized as not something to be owned only by the ruling class, but as something alive, or something present in every social action, flowing both from top to bottom and from bottom to top (Foucault, 1978).Studies done on everyday acts of resistance and social change, such as the ones by James Scott (Scott, 1987), argue that power circulates in every aspect of social life, and it is through this very circulation that the ideologies of the ruling class are reproduced.In other words, resistance to power relations was seen as an inevitable part of the power relations themselves, with individuals having no agency in a discourse (Foucault, 1978).That is, a discourse creates its own resistance, ironically within its own limits, in order to stay alive and reproduce itself.
Foucault employs the term 'panopticon' to exemplify the nature of such power relations (Foucault, 1977).Panopticon is an architectural design sketched by Jeremy Bentham for a prison.The prison in this design is made in a circular fashion.By doing so, the guards are located in a central tower, and the prisoners' cells are placed in a circle around the central tower.Therefore, the guards can monitor the prisoners all the time, while the prisoners cannot see each other.In Foucault's view, the ruling class is similarly centralized in any given society, observing each and every individual in it.Through this act of surveillance, each person in a society always feels he/she is being monitored and controlled.Therefore, they naturalize and internalize these norms and ideologies, up to a point that even in the absence of surveillance, they monitor themselves and each other, and in effect, reproduce the ideologies imposed on them through the act of surveillance and centralized control (Foucault, 1977;Kelsey & Bennett, 2014).
Until recently, Foucault's views on power and the panopticon were hugely well-received, and a great number of studies cited Foucault when attempting to delve into power relations.However, the rise of Web 2.0 has dramatically challenged all the traditional views and conceptualizations of power relations.Althusser's Ideological State Apparatus (Althusser, 2006), Frankfurt School's Culture Industry (Adorno & Horkheimer, 2001), Habermas's Publis Sphere theory (Habermas, 2006), or Foucault's Panopticon (Foucault, 1977), cannot explain the specificities of the New Media.This is a challenge for CDA theorists and other researchers working on issues of power in the New Media spheres.
The high accessibility of Web 2.0 challenges a panoptic understanding of power relations.As it is argued by Kelsey and Bennett (Kelsey & Bennett, 2014), social networks and the New Media are not panoptic and centralized at all.Rather, they are 'omnioptic' and 'synoptic', in which all individuals can monitor, challenge, and resist the ruling class, in addition to monitoring and influencing each other.That is, the circular structure of the panopticon, in which individuals/prisoners could not see the guards, does not apply anymore.With Web 2.0, it is now possible for individuals to monitor not only each other, but also keenly observe and monitor every action of the ruling class, and react to them accordingly (Kahn & Kellner, 2004;Kelsey & Bennett, 2014;McCafferty, 2011).In this sense, power is neither top-down not bottom-up anymore.Instead, it is chaotic, circulating in every direction possible, linearly and horizontally among individuals, vertically between the ruling class and the general public in both directions.This 'omnioptic' nature of power relations in the New Media (Kelsey & Bennett, 2014) is something unique to Web 2.0, for which there are no conclusive and satisfactory theories so far.This issue, differentiating the New Media from the other traditional and mainstream types, is something significant for researchers attempting to do CDA studies in these domains, since they should always consider the inefficacy of older theories of power for these new discourses.In other words, the power relations governing the New Media are not similar to those circulating in the civic life.Therefore, one cannot use the theories of communal and civic life to explain the power relations of a virtual setting (Boyd, 2008).

Flow of Information
The second theoretical issue facing CDA researchers in the New Media settings is the notion of information flow and dissemination.Again, the existing theories of information flow, such as the multistep flow of information (Katz & Lazarsfeld, 1956), cannot satisfactorily explain how information or content is produced, is disseminated, and received or used by the public.This process, putting the previous information flow models under question, is still being studied and analysed by scholars from various fields of inquiry (Bruns, 2012).
Before Web 2.0, the multi-step flow of communication could satisfactorily explain how the information would flow in a given society.The content/information was produced by the individuals or groups with sufficient levels of access, such as politicians, the media, and managers.This information was then received by opinion leaders, who had the appropriate knowledge and skills to interpret it, and then publicize it to the general public, who were the users of information (Katz & Lazarsfeld, 1956).However, the rise of Web 2.0 and its high level of accessibility has turned every individual to a producer and a user of information simultaneously.This phenomenon, labelled as 'produsage' by Axel Bruns (Bruns, 2007), has changed the older theorizations of information flow.At present, notions such as level of access are not to be seen in the traditional way, since the omnioptic nature of the New Media has given every social actor some level of access (KhosraviNik & Zia, 2014).Moreover, there is now a multitude of opinion leaders present in the social media (Bruns, 2007).Therefore, although some principles of the theory of multi-step flow of information might still work, we need more knowledge on newer forms and levels added to communication, such as the multitude of opinion leaders, higher level of access, and factors forming public interest, as well as interest publics (Boyd, 2008).
For scholars doing CDA in such settings, this is significant for two reasons.Firstly, as it was pointed out, these researchers should consider the present theories' lack of explanatory frameworks for information flow.Secondly, and perhaps more importantly, is the issue of data collection.The present models of data collection employed in CDA studies cannot address the widespread dynamics of information 'produsage'.That is, it is not enough anymore to merely collect the data from a politician's speech, a manager's discourse, or a medical professional's conversations with the patients.Rather, an array of resources should be analysed, involving both top-down discourses and bottom-up ones (Bruns & Stieglitz, 2012;Burkell, Fortier, Wong, & Simpson, 2014).Additionally, Web 2.0, the New Media, and the concept of produsage have led to the appearance of what is usually referred to as Big Data (Koteyko, 2011).In other words, the ocean of discourses available in the New Media is far from manageable and analysable given the present tools and analytical frameworks.Until recently, one could achieve an acceptable study through analysing a few pages of data.However, discourses available in the New Media consist of gigabytes of information, calling for newer modes of analysis, such as Web Analytics tools and Big Data approaches (Bruns & Liang, 2012;Bruns & Stieglitz, 2013), corpus linguistics (Baker, Gabrielatos, Khosravinik, Mcenery, & Wodak, 2008), etc.This can become clearer when one takes into account the amount of information added to the New Media every second.This huge amount cannot be analysed using the usual tools of analysis often employed in CDA studies.

Historicity
Finally, the other theoretical issue for CDA studies in the New Media is the concept of historicity, or simply the notion of time.Commonly, historicity of discourses, or their evolution and dissemination through time, was addressed either through a diachronic approach, taking into account the sociohistorical evolution of a certain discourse, or synchronically, taking into account only the present situation of a discourse.That is, in reflecting upon the relationship between the linguistic and discursive features of a discourse, the researcher would either consider how the discourse under study has become the way it is, or just focus on the discourse and the present situation of a society (Hyatt, 2005;Lehmann, 1985).
However, the New Media has added a third concept to discourses.Any given content in the New Media can be reproduced, edited, modified, or deleted at any given point in time.That is, the traditional linear concept of time cannot be applied in studies done in the New Media, since the content 'prodused' in New Media settings is 'ahistorical' (Unger, 2012).This timelessness is yet another issue challenging CDA studies in the New Media.Other than deeming previous theories insufficient, another issue created by timelessness and ahistoricity of the New Media content is the problem of 'trends'.It is often observed that a discourse, a practice, or a theme becomes 'trendy', attracting a large number of followers, 'likers', 'tweets', and 'retweets' for a limited amount of time, and then disappear as magically as it was created in the first place (Dahlgren, 2005(Dahlgren, , 2009)).This phenomenon poses two problems for scholars in the New Media, especially CDA researchers.
Firstly, as CDA is a problem-oriented discipline, it is not yet clear what criteria should be met to consider an issue a 'problem' worthy of academic analysis and discovery.That is, although trends might attract millions of followers and discourse participants at a time, and in turn, might attract a CDA researcher to do a study on them, they might die and disappear even before the study is completed.Simply put, a 'trend', as popular as it might be, cannot be considered a real social problem merely because it is popular.Such transient nature of trends in New Media makes it difficult for researchers to not waste their time on trivialities and instead identify deep-rooted problems, without being influenced by their own personal biases in choosing problems, or worse, be influenced by the biases present in 'trends' in the Social Media.
Secondly, the other problem with 'trends' is that so far, there are not enough studies done on the dynamics of them.That is, we still do not know why a topic becomes trendy, how it becomes popular and widespread, and eventually, how it loses its popularity and why it disappears (Bruns & Liang, 2012).After all, trends in the social media point to the public's interests, even if temporary and transient.The very concept of 'liking', 'following', and 're-tweeting' a certain topic should be seen as an act of social practice, and although it might not be a real 'social problem', it should be taken into consideration in understanding the dynamics of social problems and issues.The problem for CDA scholars here is a lack of knowledge in this area, which makes it difficult to reach to conclusions and acceptable interpretations of the dynamics of information flow within the New Media, and in addition to that, the connection between the New Media and 'real' or 'communal' problems and issues (Boyd, 2008).In many cases, a trend in social media is solely a social media issue, with no implications or reflections in the communal and civic life of people.A curious case of such a trend is the 'what colour is the dress' trend, haunting social media for a few months, until it suddenly disappeared and lost public interest (Klassen, 2015).Distinguishing what issues are more important and worthy of being studied is a task burdening CDA scholars in New Media circles.A CDA study in New Media settings often walks on the borderline of this contradiction.On the one hand, the problem studied might be a transient trend with no real-life implications.On the other hand, it might provide useful hints to the nature of trends and public interests in general.Therefore, the task of justifying the significance of a study has become more difficult for researchers.

CDA and New Media: Methodological Issues
Apart from the theoretical challenges facing CDA researchers in New Media, there are also a number of methodological issues that need to be taken into consideration in such studies.Methodological challenges of doing CDA in social media networks and Web 2.0 in general arise from the nature of these settings, which affects the data itself, and also some linguistic issues which should be taken into account.

Issues of data collection
So far, the small number of studies done in social media settings and on discourses of New Media from a critical discourse analytical point of view have focused on Facebook and online forums as sources of data collection (KhosraviNik & Zia, 2014).Two reasons can be thought of for such focus.Firstly, the characteristics of Facebook and online forums, and their rules of usage, allow longer texts to be published.This allows the CDA researchers to have enough analysable data.On the other hand, platforms such as Twitter, which allow only messages with 150 characters, cannot provide texts with sufficient length for in-depth linguistic analyses.The second reason for this focus is the topicrelatedness of these domains.Again, rules of usage of forums and Facebook allow like-minded users to create topics or public pages related to their issue of interest, in which they gather together and share their views.Such pages and forum topics are easy-to-access sources of data for researchers interested in doing a CDA study on a given issue.
However, collecting data from sources such as the public pages on Facebook or online forums has its own challenges and limitations.One important issue to be considered when doing such studies is 'onthe-record bias' (KhosraviNik & Zia, 2014;Khosravinik, 2014).A CDA scholar collecting data from a public Facebook page should bear in mind that a public Facebook page provides a biased discourse, since it is a gathering of like-minded people -an interest public-with similar intentions, goals, worries, ideologies, etc.This bias is significant, since it renders some tools of analysis in CDA useless.For instance, analysis of argumentation, which is one of the main tenets of the discourse-historical approach to CDA, is not useful in studies done on such discourses (KhosraviNik & Zia, 2014).As it was pointed out, the researcher is already aware of the biases shared by these pages, and argumentation analysis cannot provide anything extra.Moreover, insisting on argumentation analysis and its results as a triangulation principle is itself an act of cherry-picking in such settings, since the researcher would apply to a circular argument to confirm what is already assumed.
Another issue with the data coming from public Facebook pages and online forums is the issue of characteristics of online discourses in general.Particularly, two important features are of importance for a CDA study in an online setting.Firstly, it is shown by various studies that the 'online persona' of discourse participants might be in dire contrast with their real personality (Bullingham & Vasconcelos, 2013;Burkell et al., 2014;Stafford, 2010;Starin, Baden, Bender, Spring, & Bhattacharjee, 2008).In other words, a significant number of individuals benefit from their virtual profile, by presenting an unreal picture of themselves, and arguing from the point of view of that online persona, and not from their real self.This tendency to employ an unreal online persona is particularly significant in a CDA study, since one of the primary goals of CDA studies is providing a critique of the underlying ideologies in a discourse.However, what can be discovered in a CDA study in New Media might in no way be a real representation of the ideologies shared by 'real' individuals.That is, the results of the study should not be seen as generalizable to the offline, civic, and communal community.
Another finding of similar studies points to a rather disturbing feature of online discourses and the anonymous online persona of individuals taking part in these discourses.This feature, sometimes referred to as 'trolling tendency', points to the fact that participants active in online discourses share a feeling of anonymity and freedom.Therefore, they do not feel confined by the usual conventions of public discourse and conversations, such as politeness.This gives rise to a tendency for verbal aggression in online settings, including acts of cyber-bullying, trolling, profanity, and alike (Unger, 2012).Having this in mind, a researcher in such settings should always consider this fact that what is found in a study done only on online discourses can only be generalizable to online discourses.The challenge here is whether to confine the study only to results generalizable to a few Facebook pages, or go about the difficult task of a multi-dimensional study, collecting data from a multitude of resources, both online and offline.
Finally, as it was pointed out among the theoretical issues as well, remains the issue of the current tools employed by CDA researchers.The existing tools, such as manual collection and in-depth analysis of data, or in cases using common qualitative analysis software programs, are not enough for studies done on Big Data.In-depth manual analysis of existing data on any given topic requires years of labour.With individuals around the world producing content on any given topic every second, it is beyond the abilities of anyone to analyse even one second of Big Data manually (Bruns, 2012).Therefore, any researcher aiming to do a CDA study in online settings and New Media should look for more appropriate tools of data collection, and do not merely rely on the existing tools employed in majority of CDA studies.Sufficing to the old tools would mean that the study can in no way be satisfactorily and explanatory done.

Issues of analysis and interpretation
Apart from obstacles and limitations in the way of data collection and analysis, one other point which should be noted is the penetration ratio of the internet in various societies and geographical settings, as well as the demographics of internet users.The percentage of people using the internet is not the same in all countries and contexts.Obviously, this poses a limitation on systematic sampling of data.A CDA researcher collecting data only from online sources should always consider the penetration ratio of the internet in the society in which the study is done.Moreover, even in settings with a high penetration ratio, such as Scandinavian countries with a more than 90% rate, the demographics of internet use are still of great importance.A high penetration ratio of the internet in a country does not mean that the data collected from the internet can be generalizable to the whole population.This is because there is no direct correlation between the number of internet users in a country and the number of users following a certain topic, issue, or trend.In other words, the distinction between the public interest and interest publics should not be overlooked.For instance, although a country might have an internet penetration ratio of 95%, only a small number of the population might actively engage in a Facebook discourse on same-sex marriage.This small interest public, providing their agenda and biased views in a public Facebook page, is not representative of the public interest in the society as a whole.Therefore, a researcher working on this topic cannot in any way generalize the findings of his/her study to the whole population of that country (Khosravinik, 2014).This gap between the online and offline domains, and between interest publics and public interests, is always an issue to be considered in any studies done in the New Media, especially if the study is to take a critical stance, looking for ideologies within a discourse.

Issues of language and frameworks
It was shown how the previous models and theories of communication, public sphere, power relations, and information flow cannot be satisfactorily applied to studies done in the New Media.However, this is not the only problem in doing a CDA study on these contexts.
The difficult task of coming up with a theoretical or conceptual framework which can encompass the data under study is something that researchers often find gruesome.In many cases, especially true for less experienced researchers, the theoretical framework and methodology selected for a CDA study is one of the three main approaches to critical discourse analysis, namely the socio-dialectical approach by Fairclough (Fairclough, 1989(Fairclough, , 1995)), the discourse-historical approach by Wodak (Reisigl & Wodak, 2001, 2009), and the socio-cognitive approach of van Dijk (van Dijk, 1993, 1999).However, what is often ignored in doing so is that these approaches, and the various approaches alike, are all guidelines and general frameworks, not strict methodologies to be religiously followed (Wodak & Meyer, 2009).This is especially true when doing a study in the context of Web 2.0.
One very important consideration in selection and modification of a theoretical framework for a CDA study on social media is the language in which the data is written.Each language has its own linguistic peculiarities, including syntactic, pragmatic, and semantic features.Moreover, each setting has its own generic requirements as well.Therefore, in any CDA study, both the linguistic and generic features of the context should be considered.There are two significant aspects to be taken into account in this regard.
Firstly, the researcher should have in mind that the language used in social media has its own generic properties, and should not be approached from traditional linguistic views of language.For instance, spontaneity, grammatical mistakes, brevity, and a conversational style are features of Web 2.0 language, distinguishing it from the written register or generic structures of other media (KhosraviNik & Zia, 2014;Khosravinik, 2014).Additionally, metalinguistic features are more commonly found in the language of Web 2.0.Such features, such as emoticons, unconventional use of punctuation signs, or repetitions in the level of words and letters, are not accounted for in traditional linguistic frameworks (Eugene Agichtein Debora Donato, Aristides Gionis, Gilad Mishne, 2008;Zappavigna, 2011).On the other hand, these features carry significant semiotic aspects with them, and ignoring them damages the accountability and reliability of a study.
Another important point in selection of a theoretical framework is the issue of inter-lingual differences.It should always be noted by a researcher that the linguistic rules and interpretations found in one linguistic framework are not easily applicable to all languages.As a case in point, the concepts of modality and transitivity, which are important aspects in Systemic Functional Linguistics (SFL), and are heavily focused on in the socio-dialectical approach to CDA, are primarily based on English language.When doing a study on a language other than English, such criteria might not be applicable at all, or might need serious modifications.Therefore, blindly following SFL in studies done on languages other than English will in no way lead to reliable findings.For instance, the concept of transitivity and agency in sentences is easily applicable to English language, but might not be the same for other languages, such as German or Spanish (Aschermann, Gülzow, & Wendt, 2004;Espinoza, 1997).The same is true for a language such as Persian, which does not have the syntactic features for a sentence in which the agent is not the subject.Consider the following structural differences.In English, the syntactic structure "Jack ate the cake" could be written as either "The cake was eaten by Jack" or "The cake was eaten".Therefore, the agent of the sentence, "Jack" can be present as a subject, with more emphasis on it, or in the end of the sentence, with less emphasis, or simply removed from the sentence with no emphasis.In Systemic Functional Linguistics, these structures carry important semiotic functions.However, in a language such as Persian, there is only one acceptable syntactic structure for passive, and it is a sentence without an agent.The structure "the cake was eaten by Jack" cannot be directly translated into Persian.Such differences render SFL as not a good linguistic framework for a CDA study on Persian language.
Apart from the linguistic considerations, another aspect of doing a study on social media is the number of semiotic resources used in such domains.New Media, such as Facebook or Twitter, benefit from a combination of modalities, previously limited to certain mainstream media.For instance, sounds or voices were traditionally the only semiotic resources for the radio, or words and pictures for the press.However, multimodality is a generic feature of web spaces, in which one can find a creative combination of sound, pictures, videos, texts, and other metalingual features.This calls for theoretical frameworks which can account for the highly multimodal nature of Web 2.0 spaces.Consequently, in choosing a conceptual framework for a CDA study in such settings, researchers should not and cannot follow existing models.Doing so will lead to reducing the data to its textual properties only, and will ignore a huge set of semiotic resources.Naturally, this is not a desirable outcome for any study.

Conclusion
Critical discourse analysis, as a relatively young approach to discourse analysis, has still a long way to go.With the increasing popularity of Web 2.0 and online social media networks, more and more researchers take up the task of doing CDA studies in the New Media.We discussed the different challenges these researchers face on the way of their academic endeavours.Although there are a number of obstacles to be removed, this does not mean that studies done on these settings are not valuable (KhosraviNik & Zia, 2014).After all, there have been several instances of social change around the world, which could not succeed without Web 2.0.Incidents such as the recent Arab Uprising owe a lot to the social media (Boyd, 2008;Cottle, 2011).
The argument made here in this article is that in order to achieve a higher level of reliability and objectivity, one cannot rely only on the existing models and approaches of critical discourse analysis, simply because these approaches have been created at a time when Web 2.0 was not as popular as it is now.Therefore, the range of analytical, theoretical, and methodological tools and frameworks provided by these approaches have not predicted and accounted for the peculiarities of the New Media.As a result, a researcher attempting to do a CDA study in such spaces needs to modify these.
Certainly, with more research done on New Media, more challenges will be discovered, and in turn, more solutions will be provided.Reaching to a conclusive theorization of the New Media and all within it, such as power relations, communication models, and information dissemination, requires a collective and interdisciplinary attempt from various perspectives.CDA can benefit as much from computer science and its methodologies as from linguistics, social theory, and cultural theory.Tools developed in the field of computer science, such as Web Analytics tools, Social Media Analytics tools, sentiment analysis, data mining, etc. can all help enrich CDA studies in such circles.Moreover, newer theories in the fields of sociology and communications are emerging to address the specificities of the New Media.These theories range from semiotics to public sphere theory to theories of power.The very interdisciplinary nature of critical discourse analysis demands that researchers have a broad perspective, and employ as many tools and methods in their studies as possible.Moreover, this interdisciplinarity should also be followed in interpreting the findings and reaching to critiques of ideology.Middle ground theories might prove much more useful than the Grand theories in this regard.One should have in mind that no single theory can explain the complexity of Web 2.0, which itself is a world of its own, parallel to the civic and communal life of people.It was pointed out that the onlineoffline gap between the virtual and civic life is still an issue of debate.Certainly, critical discourse analytical studies can help a lot in filling this gap.All in all, as difficult as it is, CDA researchers in New Media need to extend the limits of their theoretical frameworks as much as possible.