Abstract: Sociological discourse analysis shares many of the procedures of other social sciences. Yet sociologists differ greatly in terms of how they approach discourse analysis, thus leading to confusion and doubts regarding the scientific status of sociological discourse analysis. In this article we attempt to clarify the methodological basis of sociological discourse analysis by differentiating it from other discourse analysis practices. To do so, we examine what sociologists actually do when they analyze discourse, while focusing on the common elements and principles shared by the different approaches to sociological discourse analysis.

Key words: discourse analysis; qualitative social research; textual analysis; contextual analysis; sociological interpretation; abduction

3.3 Sociological analysis: Discourse as information, ideology and a social product

To a large extent, sociological discourse analysis has been built up through the adoption and adaptation of methods of analysis developed in other social sciences. As a result, sociological discourse analysis shares many of the elements found in analyses conducted in a wide range of disciplines including linguistics, ethnography, anthropology and psychology, to name but a few. The peculiar manner in which discourse analysis has developed in sociology has led to an enormous diversity of styles and forms of analysis. Rather than providing a particular method for analyzing discourse from a sociological standpoint, sociologists resort to a series of practices and procedures that are used in very diverse ways in their professional practice. A brief look at the variety of manuals that have been published on this topic suffices to confirm the lack of consensus concerning what is meant by sociological discourse analysis or how to approach it¹⁾. [1]

The diversity of approaches and the lack of a formal framework for sociological discourse analysis have given rise to confusion and misconceptions among those who are familiar with this social research practice. There are at least three reasons for this. The first is that sociological discourse analysis has been identified, in a fundamental or exclusive manner, with one or another of the particular procedures associated with it. Second, sociological discourse analysis is considered a research practice lacking in rigor which depends to a lesser or greater extent on the criterion of the analyst²⁾. Finally, doubt has been cast upon sociological discourse analysis as a method of analysis in its own right. It is therefore necessary to explain what sociological discourse analysis actually involves; a task that should aim to answer two questions which basically refer to the same thing, namely what features distinguish sociological discourse analysis from analyses conducted in other scientific disciplines and what elements are shared by the different procedures in sociological discourse analysis, regardless of their apparent diversity of form? [2]

From a sociological standpoint, discourse is defined as any practice by which individuals imbue reality with meaning. When defined in these terms, discourse is found in a wide range of forms. Indeed, any social practice from a dance, ritual or a piece of music to a job contract, myth or culinary custom can be analyzed discursively³⁾. Yet the discourse of greatest interest to sociologists is that which takes a verbal form, be it written or spoken. The reason for this special interest in verbal discourse is twofold: a practical one and a theoretical one. In practice, verbal discourse is discourse that can be accessed and examined by the analyst. Indeed, analyses of other forms of discourse, for example visual discourse, often rest on translating the discourse into a verbalized format by means of detailed descriptions. In theory, verbal discourse is a privileged means of producing and transmitting meaning. Although visual discourse, and to a lesser degree harmonic and spatial discourse, is becoming increasingly widespread, verbal communication remains the most common way of producing and transmitting meaning in our society. [3]

Another basic distinction lies in the difference between spontaneous discourse and induced discourse. Spontaneous discourse refers to discourse produced by subjects in their everyday lives. Books, records of court proceedings or television programs, for example, frequently comprise the basic material of sociological discourse analysis. Although these discourses are produced by subjects for specific aims and these aims differ from those of sociologists, they are appropriate for certain types of research purposes. However, induced discourses, produced within the framework of research, are more often the fundamental material which sociologists work with when conducting analyses. In the majority of cases, sociologists prefer to focus on discourse induced by social research methods as it enables them to maintain a relatively high level of control over the conditions in which these discourses emerge. This type of discourse is usually produced in the framework of in-depth interviews or via group dynamics, particularly in the form of group discussions. For this reason, we will refer chiefly to induced discourse when examining the methods used in sociological discourse analysis, while bearing in mind that these approaches can also be applied to spontaneous discourse. [4]

The interest in discourse as a means of understanding social reality is based on the notion of the subjective orientation of social action. Given that social action is guided by the meaning that individuals attach to their actions, we must account for this meaning when attempting to understand and explain the action. Yet meaning is not only a product of individual constraints and beliefs. Instead, the meanings that guide individual actions are, to a large degree, socially produced and shared patterns. In this sense, Alfred SCHUTZ highlights the need to account for the individual's viewpoint in order to explain social action. He also indicates the importance of intersubjectivity as an essential element in the structure of the commonsense world (SCHUTZ, 1962, pp.3ff.; 1964, pp.3ff.). The world in which individuals orient or project their actions is known and understood by them to be a socially organized world. Hence what I know and understand about this world coincides, to some degree, with the knowledge and understanding of the others with whom I relate. As SCHUTZ explains:

"... from the beginning this orientation through understanding occurs in cooperation with other human beings: this world has meaning not only for me but also for you and you and everyone. My experience of the world justifies and corrects itself by the experience of the others with whom I am interrelated by common knowledge, common works, and common suffering. The world, interpreted as the possible field of action for us all: that is the first and most primitive principle of organization of my knowledge of the exterior world in general" (SCHUTZ, 1964, p.9). [5]

For SCHUTZ, the goal of the social sciences should be to provide an explanation of social action based on this subjective viewpoint. In his opinion, the problem of the social sciences lies in how to obtain objective knowledge about this subjective reality. To this end, he proposes the construction of ideal types as a method for gaining scientific knowledge about subjectivity. Thus although he emphasizes the importance of intersubjectivity in the making of this subjective viewpoint, he believes that the substantive research of intersubjective knowledge and understanding (systems of significances, socially shared types) is not relevant to the scientific exploration of subjectivity. However, in contrast to this, qualitative social research aims to obtain objective knowledge about subjectivity from intersubjectivity. [6]

Gaining objective knowledge from intersubjectivity leads to a series of methodological problems that differ from those analysed by SCHUTZ with respect to obtaining objective knowledge from subjectivity. Given that intersubjectivity is a required element of social interaction, it leaves an imprint on the outcome of such interaction, particularly on the discourse that is the product of communication. As SCHUTZ states:

"successful communication is possible only between persons, social groups, nations, etc., who share a substantially similar system of relevances. The greater the differences between their systems of relevances, the fewer the chances for the success of the communication. Complete disparity of the systems of relevances makes the establishment of a universe of discourse entirely impossible" (SCHUTZ, 1962, p.323). [7]

If communication is possible only in the framework of intersubjectivity, its outcome or product, that is, discourse, embodies this intersubjectivity in an implicit manner and can thus be explained through analysis. Qualitative social research focuses its attention on this dimension of social life insofar as the content, extent, limits and structure of intersubjectivity constitute fundamental elements of the subjective orientation of social action. [8]

If we bear in mind that the social universe is largely a space of shared meaning, then discursive practices are clearly important for our knowledge and understanding of social reality. Discourse analysis as a social research method is therefore grounded in two basic assumptions: 1) the knowledge of social intersubjectivity provides us with indirect knowledge about the social order because intersubjectivity is a product of it and because the social order is formed and functions through this social intersubjectivity; 2) discourse analysis allows us to understand social intersubjectivity because discourses contain it and because social intersubjectivity is produced through discursive practices. [9]

In order to interpret discourse from a sociological standpoint, discourse must first be analyzed from both a textual and a contextual approach. There are, therefore, three different levels of analysis: a textual level, a contextual level and an interpretive level. Although text- and context-based analyses are elements of sociological discourse analysis, they are not in themselves sociological analyses. Textual analysis allows us to characterize discourse as it focuses chiefly on the utterance and considers the discourse as an object of study. Contextual analysis, on the other hand, allows us to understand discourse as it centers on the enunciation, considering the discourse as a singular act or event. Finally, interpretation provides an explanation of the discourse as it addresses sociological aspects and considers discourse as information, ideology or a social product⁴⁾. [10]

These three levels could be considered as a linear process that moves from textual analysis to contextual analysis and finally to interpretation; the latter is understood as the ultimate aim of the analysis. However, this is only so in part. While it is true that there is a principal line of analysis which goes from textual and contextual analysis to interpretation, in practice these three levels do not constitute three separate stages or moments of analysis. Instead, it is common for the analysis to be carried out simultaneously on all three levels in a backwards and forwards movement that resembles a continuous dialogue among the levels. It is therefore not a linear process, but one which is circular and bidirectional and only concludes when the analyst considers that the research objectives have been achieved. Textual analyses and contextual analyses give rise to sociological interpretations, which are, in turn, present in a more or less implicit manner at both levels since it is precisely these interpretations that are of interest or value to the sociological analysis. Textual analysis involves contextual analysis in that it requires contextualization, while contextual analyses orient new textual analyses. And all of this takes place in a circular, on-going process in which the different types of analyses feed back into one another (Diagram 1). [11]

In what follows, we will address these three levels of analysis separately and provide a brief overview of the most relevant methods and procedures of analysis for each. However, we must bear in mind that, in analytical practice, numerous relations arise, merge and intertwine between these levels, making it difficult on occasion to assign a specific type of analysis to a given level. For purposes of clarity, however, we have assigned an analytical procedure to the levels based on the particular characteristics of each.

At an initial stage, discourse analysis centers on textuality. The relationship between discourse and text is not univocal, thus the two concepts should not be confused or equated. Indeed, every piece of discourse has a textual form or can acquire it; the same text may include different discourses or the same discourse may adopt different textual forms. Textual analysis considers discourse as an object, giving it the appearance of objectivity and making it especially interesting for those who approach discourse analysis from positivist scientific positions. But this objectivity is only so in appearance because when discourse is considered exclusively as an object of study, the analyst is not eliminated, but hidden. Indeed, behind the seeming objectivity of textual analyses, there is at least a subject who reads the texts, selects the relevant elements and establishes the pertinent relations or significances. Textual analysis can therefore be viewed as a level of discourse analysis in which the subject-analyst is shielded behind standardised methods, allowing her or him to go unnoticed. On the other hand, while it is true that discourses have an objective dimension, they are not only objects. From a sociological standpoint, therefore, the study of the objective dimension of discourse is solely one phase or level of analysis. Discourse not only embeds meaning but also produces it. It is an act and a product as well as an object: for this reason the first level of "objective" analysis is clearly insufficient. In this regard, the confusion between text and discourse can only be explained as an attempt to approach textual analysis in a totalizing manner. [13]

Very few discourses of interest to sociologists appear in a textual form in an immediate manner. Only documents and publications (books, journals or newspapers) contain primary textualized discourse. Thus the first step that is normally taken in textual analysis is to translate the discourse into a textual form. This translation of non-textual discourse to a textual form constitutes the first phase of textual analysis and should therefore be done according to rigorous criteria and procedures. To do so, two procedures are used: description, which is applied to non-verbal discourse, and transcription, which is applied to spoken discourse. The fundamental criterion for translation of both sorts is that it be done in a literal and detailed manner so as to recover all the nuances of the discourse in the best possible manner. It is important to emphasize that the translation of discourse into a textual form is not only important for the first level of analysis, but is also fundamental for contextual analysis and for the interpretation of discourse. Thus, both description and transcription should include all the antecedents and contextual elements of the text that can contribute to its interpretation. So, transcription should include all the non-verbal events (moments of silence and their duration, modulations, emphasis, meaningful gestures and expressions, etc.) as well as the verbal events, whilst the description of a dance or ritual, which should be equally as detailed, should include all of the elements of the context in which the discourse has been constructed. [14]

Textual analysis involves characterizing or determining the composition and structure of the discourse. The aim of textual analysis is not to provide a reduced version of the discourse in order to facilitate study. On the contrary, textual discourse analysis more closely resembles an ensemble approach whereby information is enlarged upon and multiplied rather than reduced. To do so, sociologists usually resort to two techniques: content analysis and semiotic analysis. Indeed, the main schools or trends in textual discourse analysis have been founded upon these two methods of analysis. [15]

For our purposes, these two approaches to analysis are not in the least incompatible or in opposition to one another, but provide very valuable material for the textual characterization of discourse. The choice of approaches will depend solely on the specific aims of the researcher. The misuse or abuse of one or the other is usually the result of a hegemonic, if not exclusive, method of analysis. [16]

Content analysis mainly consists of breaking down or fragmenting the text into pertinent units of information for their subsequent coding and categorization. Content analysis is often considered a strictly inductive method and is even referred to as a process of theory construction. However, the entire procedure is governed by theoretically established categories: the interest or value of the text, how to break it down and, most importantly, how to classify the fragments depends on the theoretical aims of the researcher. Although it is true that the initially established system of categories can be enriched through the analysis, the notion that the text analyses itself or sets down the conditions for analysis is but a mere illusion. [17]

Once the text has been broken down and coded, different methods of analysis are used⁵⁾. Originally, content analysis was markedly quantitative in nature as it centered on the manifest content of messages and was largely limited to a descriptive aim. Indeed, one of the classic definitions of content analysis considers the method to be "a research technique for the objective, systematic-quantitative description of the manifest content of communication" (BERELSON, 1952, p.18). This quantitative-based orientation has accompanied content analysis to the present day, albeit the multivariate analysis methods used have become increasingly sophisticated (multiple correspondence analysis, factor analysis, etc.). [18]

Different types of content analysis can be used depending on the objectives pursued by the researcher or the characteristics of the texts to be analyzed (e.g., the analysis of spontaneous discourse such as newspaper headlines). One type of content analysis that is especially interesting for discourse produced in the context of sociological research (induced discourse) is thematic analysis. This type of analysis centers on the themes or topics around which the discourse is developed. The selection of pertinent topics, the order in which they appear, the time dedicated to each, the relationships between the different topics or how they emerge (in a spontaneous or suggested way) are very important questions to bear in mind when characterizing discourse. [19]

In the 1960s, however, scholars began to point to the need for qualitative approaches to content analysis. One of the most important contributions in this regard is what is known as grounded theory. Originally developed by GLASER and STRAUSS⁶⁾, the basic aim of these qualitative approaches has been to recover the latent meaning of discourse in content analysis. Thus, in addition to what discourses say (manifest content), it is also necessary to account for what is suggested by them or even what is hidden in them⁷⁾. Moreover, the constant comparative method (CCM) derived from grounded theory emphasizes the need for content analysis to pay greater attention to the textual structure (analysis of semantic networks, hierarchical trees, intensity analysis, etc.)⁸⁾. Thus we can speak of a growing interest in the textual structure within content analysis and within discourse analysis in general⁹⁾. [20]

Today, computer applications used in discourse analysis have largely facilitated this type of analysis, leading to greater precision and refinement. While the usefulness of these computer applications chiefly stems from the fact that they adapt well to content analysis, they are also useful for storing data and, in general, for making information more manageable; advantages that are particularly important when dealing with a large amount of information. But in addition to these unquestionable advantages, the use of computer applications for content analysis also has its shortcomings. Perhaps the most important of these is that they give rise to a mechanized notion of analysis by which analysis follows its own logic without the intervention of the subject/researcher. This idea of an "agentless" analysis proves to be very appealing to those interested in a strictly objective analysis. In our opinion, however, the use of computer programs does not eliminate the need for the subject/researcher but it does serve to conceal the researcher as an agent in the analysis process beyond the role of program executor. Although this does not eliminate the necessary intervention of the subject/researcher, it prevents this intervention from being questioned, thus permitting greater maneuverability. Objectivity, however, can only be achieved by taking into account the intervention of the subject, thus making the intervention explicit and subject to criticism. For this reason, the mechanization of the research task must not be used as an analytical shortcut, as it alone does not ensure objectivity. [21]

Content analysis takes meaning for granted as it is based on the assumption that there is a community of meaning or set of shared meanings (language) which determine the meaning of the discourse in an immediate and problem-free manner. In contrast, semiotic analysis does not negate the importance of these shared meanings, but problematizes them: the meaning of discourse is not determined by language or at least not in an absolute and definitive manner. There is not a hierarchical or programmed relationship between language and speech (discourse), but rather a mutually determined dialectical relationship insofar as discourses use language (shared meanings) as a means of expression, but in doing so they also modify or renew it. [22]

Within the field of semiotic analysis it is also possible to distinguish between two broad types or trends: structural semiotic analysis and formal semiotic analysis. Structural semiotic analysis attempts to reveal hidden linguistic codes in order to discover and describe their internal logic, which is understood as a generative matrix that "reproduces" the text¹⁰⁾. This type of structural discourse analysis was largely developed and gained widespread acceptance in the 1960s and 1970s. However, since the end of the past century, the structural analysis of texts has been widely criticized and questioned. One of the fundamental criticisms of this approach has to do with the fact that it views the textual structure of discourses as being autonomous from and external to the subjects that produce them. Indeed, some of the basic assumptions of structural text analysis are clearly abusive and consider textual structures in a totalizing manner; a position which is then imposed upon the discursive practices of the subjects. According to these extreme positions, subjects merely update pre-existing discursive structures which are reproduced outside of their discursive activity¹¹⁾. However, this does not minimize the interest and usefulness of structural analysis for sociological discourse analysis, provided it is considered but another tool to be used in textual analysis and does not supplant the sociological interpretation of discourse, which according to our approach, occurs at a different level. In other words, the structural semiotic analysis of texts is a very useful tool for sociologists when it is not used in a totalizing manner, that is, provided that it is restricted to the textual level and does not lead to sociologically unwarranted interpretations. [23]

Certain poststructuralist approaches do, however, merit stronger criticism insofar as they constitute a nihilist reversal of structuralism and the rejection of both the textual logic and true references of discourse, thus eschewing the existence of social structures (ALONSO, 1988). For sociological analysis, deconstruction—the method preferred by poststructuralists—is little more than a game of signifiers; a game of intra and intertextual differences that may be lots of fun to play, but contribute little to the analysis. They contribute little to sociological discourse analysis because deconstruction attempts to demonstrate that discourse is not transcendent; an aim that is contrary to that pursued by sociologists, that is, to provide evidence for and demonstrate the connections between social discourses and the social reality in which they are produced and in which they circulate. [24]

Formal semiotic analysis, on the other hand, centers its attention on the effects of the meaning of discourse at the enunciation level. It is therefore a first step for considering the context in which the discourse is produced and in which it acts. In the text, form is as significant as content in terms of producing meaning. The formal analysis of a text involves accounting for the rhetorical figures it contains: the types of deixis used (I, you, us, here, there, tomorrow…), verb tenses and modals to indicate doubt, requests or certainty, among others. As we will see below, the difference between contextual analysis and formal semiotic analysis lies in the fact that in the latter these rhetorical elements "rather than referring to a real enunciation, give some indication about the type of communication established, they define the framework of communication" (LOZANO & PEÑA-MARÍN, 1988, pp.295-296). Other relevant questions in terms of the form of texts are the use of lexis, rhetorical devices (metaphors and metonymy) and syntactic forms as they are mechanisms for producing, constraining or liberating meaning. [25]

Yet a textual analysis that makes use of all of the above procedures (content analysis, structural semiotics and formal semiotics) would involve enormous effort, particularly if the analysis is conducted in an exhaustive manner and the number of texts to be analyzed is large. More so if we bear in mind that textual analysis is only the first step—albeit a very important one—in the process of sociological discourse analysis proposed here. For this reason, it is rare for analysts to use each and every one of the available methods. Indeed, given that textual analysis is the first stage of sociological analysis, analysts normally prefer to use only one of the procedures and solely resort to the others in a partial manner to gain deeper insight into a specific aspect of the text. Although a full array of tools are available, it is up to the analyst to select and use them depending on his particular research objectives, the resources available to him (particularly with regard to time) or even his own preferences or theoretical orientations. In general, however, it is advisable to use several methods of textual analysis, albeit as we have said, to a differing degree. By using a diversity of methods, the analyst will be able to gain a broader perspective and contrast a variety of elements, thus enriching the analysis. [26]

The second level of sociological discourse analysis centers on context. Context is understood as the space in which the discourse has emerged and in which it acquires meaning. On this level, discourse is understood as a singular event produced by subjects who are immersed in a specific time and place within a given symbolic universe and who have their own discursive intentions. Accordingly, it is possible to make a distinction between two types of contexts: situational contexts and intertextual contexts, giving rise in turn to two types of analysis: situational analysis and intertextual analysis¹²⁾. [27]

Situational discourse analysis requires a detailed description of the circumstances in which the discourse has been produced and the characteristics of the subjects that produce it. For example, with regard to induced discourse (discourse produced in the context of social research), if the discourse is individual or collective, if a prior relationship has existed between the subjects researched or between them and the researcher, the available resources (time, discursive capacity and discretion) and even the comfort and habitability of the space are, among others, relevant questions for understanding the local meaning of the discourse. [28]

At this point, the analysis centers on the more pragmatic aspects of the discourse. The basic assumption is that discourse has an intentional dimension and the analyst must therefore inquire as to why the discourse has been produced and for what aim. Situational analysis thus goes beyond a mere description of discourse to provide an initial explanation at a micro-sociological level. Situational analysis requires having sufficient information and an adequate understanding of the circumstances in which the discourse is produced, but also, and more importantly, it focuses on the interactions and dialogical processes involved in its production. While who produced it, under what circumstances and with what purpose are relevant questions for understanding the local meaning of discourse, it is also essential to determine how it was produced, that is, what social processes played a role in producing it. As regards induced discourse, one of the most important questions to be analyzed is the role of the researcher and especially the relationship between the researcher and the subject or subjects involved. Due to the complex nature of situational discourse analysis, the sociologist must resort to different procedures, among them the analysis of discourse positions, frame analysis and conversation analysis. [29]

The analysis of discourse positions is a first step to linking specific discourses with the social space in which they have emerged and is therefore an initial approach to the sociological interpretation of discourse. Discourse positions are understood as typical, socially defined discursive roles that subjects adopt in their concrete discursive practices. But contextual analysis is not as interested in the more or less generalized nature of these positions as it is in the discursive strategies adopted by subjects¹³⁾. When understood as such, discourse positions permit researchers to reconstruct communicative interactions through which the discourse has been produced and in this way gain a better understanding of their meaning from the viewpoint of the subjects that take part in them. [30]

Frame analysis, on the other hand, is a very useful procedure for situational discourse analysis. It is a form of analysis drawn from GOFFMAN (1986), who adopted the concept of frame originally formulated by BATESON in order to extend upon and integrate the notions of façade, performance, frontstage/backstage, role and role-distance (HERRERA & SORIANO, 2004). Frame analysis holds that the local norms governing everyday interactions must be accounted for in order to understand and explain social action. Understanding discourse as a product of communicative interaction therefore requires accounting for the norms that govern the concrete situations in which discourse is produced. These norms are very diverse and both explicit and implicit, ranging from formal codes of conduct to conventions that are shared to some degree by those engaging in the communication. [31]

The general framework of the communicative exchange in which discourse is induced in a social research setting through interviews or group discussions is established by explicit norms set down by the researcher and accepted by the subjects¹⁴⁾. Nonetheless, the acceptance of these interview or group dynamics is often limited. Subjects may understand what is asked of them in an imprecise and erroneous manner, they may openly question the researcher's authority or they may propose other norms depending on their own particular interests or definition of the situation¹⁵⁾. These considerations regarding the dialogic nature of discourse production, which may occur to a lesser or greater degree depending on the circumstances in which discourse is produced¹⁶⁾, open situational analysis to the possibility of negotiating the meaning of the situation itself. In other words, the implicit and explicit norms that govern the communicative event in which discourse is produced are insufficient for characterizing the situation in which such an exchange has occurred. Hence, they are also insufficient for understanding the meaning that the discourse has for the subjects involved. The possibility of negotiating the situation and the norms that govern it (and, in a broader sense, the very meaning of the discourse) means that we must turn to conversational analysis. As we said above, this is the third procedure used to analyze the situational context of discourse. [32]

Conversation analysis views everyday communicative events as a process of negotiating meaning. What is negotiated is the meaning of the communicative situation itself and with it, the meaning of the discourse that is produced. This type of analysis focuses on the pragmatic component of language: through language subjects not only say things, but also do things. And one of the most important things that subjects do with language is to define the situations in which they are immersed. This negotiation of the meaning of a situation is of crucial importance for understanding discourse as it permits us to determine how those involved interpret the communication and its product. Conversational analysis problematizes a question that is evident in our everyday lives: what do speakers want to say when they communicate with us. Negotiating the meaning of the situations we engage in in our everyday lives is a process that goes practically unnoticed. Only when there is serious disagreement as to the meaning of a situation do we use explicit mechanisms to repair it (for example, by asking for or offering explanations) or we choose to put an end to the communication. However, when sociologists analyze discourse they should not take for granted the processes of communicative alignment that occurs among speakers. On the one hand, these processes have a very important substantive value for our analysis, in that they enable us to determine what is being "talked" about; while on the other, they are an element of the communicative situation that is fundamental to our understanding of what the subjects want to say¹⁷⁾. [33]

Being attentive to the discursive roles adopted by speakers (the analysis of discourse positions), the implicit and explicit norms that govern the communicative events in which discourse is produced (frame analysis) and the processes of negotiation in a discursive situation (conversation analysis), enables us to characterize the situational context of the discourse and thus gain deeper insight into what it means for the subjects that produce it. But the context of discourse is not only situational; it is also, as we said before, intertextual. All discourse is embedded in a symbolic and cultural universe in which it acquires meaning. Thus intertextual analysis permits us to understand discourse by referring to all of the discourses that circulate in the social space. [34]

Two forms of intertextual analyses are especially interesting for sociologists who analyze discourse. The first of these was originally put forward by Norman FAIRCLOUGH (1995). This approach consists of seeking the presence of features from other discourses in the discourse to be analyzed. This concept of intertextuality is built on the notion that subjects resort to discourses circulating in the social space in order to produce their own discourse. Discursive activity thus understood is equivalent to selecting and combining elements from other discourses. Intertextuality, which is understood as a discursive bricolage, led FAIRCLOUGH to interpret discourse as a symptom of ideological domination: subjects are reduced to being mere reproducers of dominant discourses¹⁸⁾. An interpretation of this type clearly limits the interest that this approach holds for sociological discourse analysis. Of greater interest is the concept of intertextuality proposed by FOUCAULT (1973). Rather than identifying external discourses, this method is based on comparative analysis: the meaning of discourse emerges in reference to other discourses with which it engages in dialogue, be it in an explicit or implicit manner. The analyst must ask "each fragment of an analyzed discourse about its presuppositions, which other discourses it dialogues with and thus with which other discourse or discourses it has an associative or conflictive relationship" (ALONSO & CALLEJO, 1999, p.49). The value of discourse therefore stems from its similarities and differences with respect to other discourses. [35]

The contextual analysis described here is undoubtedly of enormous interest in itself. As highlighted above, this type of analysis often attempts to provide an explanation for the communicative processes that occur in everyday interactions, thus offering scientific insight into one of the most important processes which, from a micro-sociological view, governs social life. However, from the standpoint of sociological analysis, interest in contextual analysis is merely instrumental. The contextual level of analysis permits us to understand the meaning that discourse has for those who engage in it and therefore centers on how the subjects involved interpret the social situations in which the discourses emerge and in the discursive spheres in which they are projected. Sociological discourse analysis must account for the subjects' interpretations of the event. That is, it is necessary to understand the meaning that the discourse holds for them, but only to formulate one's own interpretation; an interpretation which must be compatible with those interpretations but is not directly derivable from them. [36]

3.3 Sociological analysis: Discourse as information, ideology and a social product

At the final level of sociological analysis, discourse requires interpretation. Yet while interpretation constitutes a third level of sociological discourse analysis, it is also present throughout the analytical process, that is, in the two prior levels. The establishment, for example, of a system of categories for content analysis or the textual structure of the discourse already involves a certain process of interpretation. As stated above, this is so because although interpretation is the final level of analysis, and as such the culmination of the sociological analysis, analysis is conducted in a constant and bidirectional manner among these three levels. [37]

The sociological interpretation of discourse involves making connections between the discourses analyzed and the social space in which they have emerged. These links or connections can be very diverse depending on the analyst's own theoretical orientation. In practice, however, sociological interpretations of discourse are limited to three types: those which consider discourse as social information, those which consider discourse as a reflection of the ideologies of the subjects who engage in it, and those which consider discourse as a social product. Yet these three types of interpretations are not mutually exclusive. Indeed, analysts often use a combination of two or even all three forms of interpretation. [38]

A first type of sociological interpretation focuses on the informative dimension of discourse. The mere fact that subjects are involved in, and have contact with, the social reality under investigation means that they are familiar with and knowledgeable about it. Discourses contain this knowledge of social reality; hence the analysis should provide relevant information about it. This type of interpretation attempts to explain discourse in terms of the social competence of subjects as informants, namely their knowledge of the reality, their expository capacity, etc. [39]

The quality of information about the social reality that is contained in discourse varies according to the level of knowledge that each individual has about this social reality. But this quality is limited since the information that subjects have about social reality is partial in a twofold way. First because it refers to a limited parcel of this reality—the parcel subjects are in contact with and their degree of contact depending on the position they occupy within the social structure. However, this limitation can be overcome by resorting to a range of informants who can offer a global vision of the reality that is of interest to us. But the information that subjects provide about this social reality is also partial insofar as it is filtered by their own particular point of view. In addition to the informative component, discourses include an ideological component. To put it another way, discourses contain knowledge about the social reality, but this reality is perceived from the viewpoint of the subjects that engage in them. Nonetheless, analysts can get around this limitation, or at least mitigate it, by abstracting in the interpretation all of the aspects of the discourse that are attributable to the subjective position of the informants. [40]

In spite of these limitations, it is common practice to interpret discourse as information and a very useful one for the purpose of sociological analysis. Indeed, this type of interpretation is prevalent in analyses based on the grounded theory approach or the analysis of expert discourses in applied research. The reason for this widespread interest in the informative interpretation of discourse can be sought in its usefulness since, in practice, social discourse analysis provides us with valid and relevant information about the social reality. In the following section we will return to this important question when discussing inductive inference as the logic upon which these types of interpretation are based. [41]

In contrast, the ideological partiality of discourse, which is a limitation to its informative interpretation, is the basis for interpreting discourse as ideology¹⁹⁾. What is of interest to the analyst in this type of interpretation is the subject's particular viewpoint. This viewpoint is not considered to be a subjectivist bias of the discourse but an indication of ideological constructs, which are understood as intersubjective modes of perceiving the world and finding one's place in it; a process common to subjects immersed in concrete social and historical contexts. This type of ideological interpretation is a characteristic feature of critical discourse analysis (CDA), which aims to demonstrate how social discourses are impregnated by dominant discourses projected from sources of power (VAN DIJK, 1999). Discourse is therefore understood to mirror mechanisms of ideological domination. But discourse can also be considered a potential mechanism of liberation. Discourse in this case is produced by the critical analyst who reveals or manifests these mechanisms of ideological domination in an attempt to overcome or eliminate them. [42]

CDA has grown out of approaches more akin to psychology or social psychology than sociology, although this does not mean that it is lacking in interest for sociological interpretation. Indeed, the presence of diverse mental constructs such as shared patterns of understanding and interpretation, interpretative repertoires (POTTER & WETHERELL, 1987) or mental representations can be derived from discourse analysis. In an explicit manner, BILLIG (1991) refers to these mental constructs as ideologies. This type of interpretation places emphasis on the cognitive structures involved in discourse where these structures are understood as shared patterns of meaning or common ways of perceiving reality. Nonetheless, the sociological interest of CDA is limited in that it considers the pragmatic effects of discourse in relation to the immediate social context, but does not link it to the broader social context. [43]

The ideological interpretation of discourse is also characteristic of Pierre BOURDIEU's analysis of what he terms linguistic markets (BOURDIEU, 1991). According to BOURDIEU, discourse reflects the habitus of the subject who produces it. In this sense, habitus is understood as the discursive competence of the subject, which derives from belonging to a given social group and from the social experience that is conditioned by this belonging. Social discourses will not only vary, in the sense that they will depend on the social position of the individual who engages in them, but they will also have a different social value. This led BOURDIEU to speak of linguistic markets as mechanisms that establish and maintain the unequal value of different social discourses. The diversity of social discourses is therefore considered a reflection of social inequality and a cultural mechanism of domination or a means to preserve these social inequalities (ALONSO, 2002). [44]

A third type of sociological interpretation considers discourse as a social product. Every product reflects the social conditions under which it has been produced. By analyzing the product, fundamental aspects of life and the social structure are revealed to us in an indirect manner. If this is so for any product, it should hold even more so for discourse in that discourse is a product that carries a heavy symbolic load. The key questions for this type of interpretation are: Why have certain discourses been produced (and not others)? What social conditions have allowed certain discourses to emerge and not others? This type of interpretation involves taking a larger step or breaking away from discourse in that it establishes a connection with the wider social context. [45]

The interpretation of discourse as a social product is present to a greater or lesser degree in practically all sociological discourse analyses. In the sphere of Spanish sociology, the interpretation of discourse as a social product is a salient feature of the classic approaches of Jesus IBÁÑEZ (1979, 1985) or the more recent approaches of Fernando CONDE (2002, 2007). It is also characteristic of Foucaultian Analysis, although this particular approach is usually combined with an ideological interpretation.

"The starting point of the FOUCAULTian analysis of statements is thus the diversity of all statements whose positivity is in need of investigation. The point here is to analyse the historical conditions of the actual existence of statements. (...) First he asks which object or area of knowledge is discursively produced; second, he asks according to what logic is the terminology constructed; third, he asks who authorized it; and finally, he asks which strategic goals are being pursued in the discourse" (Díaz-Bone et al., 2007, p.5). [46]

Unquestionably, interpretation is the aspect of sociological discourse analysis that has aroused greatest suspicion. This is because interpretation requires moving beyond the specific discourse being analyzed. But although a leap must be made when interpreting discourse, it is not a leap in the dark. On the one hand, it is grounded in textual and contextual analyses: the information about the discourse produced in the two previous levels provides a strong foothold for the interpretative leap. On the other hand, although analysts have plenty of room to manoeuvre in sociological discourse analysis, their analyses are conducted according to a strict scientific logic which will be discussed in further detail below.

The sociological interpretation of discourse is based on a logic that is uncommon or at least different from the logic followed in the majority of scientific inferences. On occasion, this has led to the view that the interpretations are poorly founded or even arbitrary. With a view to correcting these misunderstandings and misconceptions, we will discuss the logic or logics used in this research practice; specifically inductive logic, abductive logic or a combination of both. [48]

At times sociological interpretations of discourse are presented in the form of inductive inferences insofar as they are generalizations based on observations²⁰⁾. These generalizations, however, have some peculiar characteristics in the case of sociological discourse interpretation, particularly with regard to the number of cases analysts work with. Although a large number of cases allow inductions to be verified with a greater degree of certainty, sociological interpretation of discourse does not require a large number of cases to make inductive inferences. In fact, it is possible to make inductive inferences with a small number of discourses in that society is a complex system. The different elements of such systems are not isolated from one another, but are intertwined with other elements of the system in such a manner that the information they possess about it is directly derived from the position they occupy within the system. Hence the information one individual provides is interchangeable with the information provided by any other individual in the same or a similar social position. It is therefore sufficient to examine a small sample of discourses produced by subjects who occupy positions that are significant to the research inquiry²¹⁾. [49]

The peculiar form adopted by induction in the sociological interpretation of discourse has important consequences in terms of how unexpected or unforeseen cases are dealt with. Contrary to what POPPER (1965) claims, when the evidence does not fit into the theory, we must not necessarily abandon or refute the theory. Instead, unforeseeable results should lead us to modify and progressively refine our theories to explain these new findings. Thus, new discoveries do not always question the validity of what is already known, but enable us to enlarge upon our knowledge. [50]

While unexpected or unforeseeable results do not always lead us to refute the theoretical framework upon which our predictions are based, on occasion they do. When unexpected results cannot be reincorporated by extending the theory, we must abandon it and seek a new theory that serves to explain the diversity of what is real²²⁾. This search for a new theory brings us to the second logic-based approach to sociological discourse interpretation: abduction. Although interpretation by induction is a very frequent and fruitful practice, interpretation by abduction is the major contribution of sociological discourse analysis (ALONSO, 1998). Abduction can be defined as an inference in which the conclusion is a hypothesis. The term was originally defined by PEIRCE, who held that:

"accepting the conclusion that an explanation is needed when facts contrary to what we should expect emerge, it follows that the explanation must be such a proposition as would lead to the prediction of the observed facts, either as necessary consequences or at least as very probable under the circumstances. A hypothesis, then, has to be adopted, which is likely in itself, and renders the facts likely. This step of adopting a hypothesis as being suggested by the facts, is what I call abduction" (PEIRCE, 1901, p.202)²³⁾. [51]

Clearly, this is a weak form of inference. This weakness, however, is not a problem for PEIRCE, who never questioned the logical nature of abductive inference²⁴⁾. Indeed, one of PEIRCE's main contributions was to demonstrate that reasoning is not limited solely to deduction, but also involves induction and abduction (DEBROCK, 1998). Moreover, abductive inference is of special importance to scientific method in that it is the only process by which new ideas can be introduced in science and is therefore the logical basis of scientific creativity. Hence abduction, as well as deduction and induction, are processes of inference or reasoning that constitute three interdependent states of scientific research. Scientific research starts from abductively inferred hypotheses, is followed by deductively inferred implications of those hypotheses and concludes with the inductively inferred empirical verification of those implications. While induction (and deduction) responds to a logic of scientific verification, abduction responds to a logic of scientific discovery (HOFFMAN, 1998)²⁵⁾. [52]

This weakness is not, however, the main problem of abductive inference. The real problem lies in how to formulate an abduction. PEIRCE was not especially clear on this point when he referred to a "flash of understanding" or when attributing abductive capacity to an adaptive instinct: abduction emerges from a need (the need to explain surprising or unexpected facts) and depends on a capacity developed by human beings, particularly scientists. Yet making the formulation of abductions depend on an instinct would seem to contradict the very nature of logical inference. However, as PEIRCE did not believe a formal procedure was necessary to formulate abductions, he did not consider this question a problem. In fact, later attempts to formalize abduction have not obtained very promising results. The formalization of scientific creativity is not only difficult and even counterproductive, but in PEIRCE's opinion, is unnecessary as it does not compromise the logical nature of science. [53]

Although PEIRCE does not establish procedures to formulate abductions, he does set down criteria to distinguish between a good and a bad abduction; a sort of pragmatic guide for their formulation. Specifically, he provides three criteria: the need for abduction to propose truly "new" ideas or explanations, the need to derive empirically contrastable predictions from the hypotheses and the need for the hypotheses to fit in with or give an adequate account of the social and historical context in which they emerge. The first of these conditions cautions us against false abductions, that is, those which, in the strict sense, are a veiled deduction in that they are based on an analogy of properties (DEBROCK, 1998). The second condition focuses on the role of abduction within the process of scientific research: for abduction to be a driving force of scientific research, it must permit derivation through the deduction of empirically contrastable predictions (SANTAELLA, 1998). Finally, the third condition alludes to scientific intersubjectivity as a criterion that permits the set of possible abductions to be established (HOFFMAN, 1998). [54]

Because sociological interpretations of discourse adopt the logical form of an abduction, they provide an explanation of the discourse as an indication or symptom of broader social phenomena. In this sense, abduction is akin to detective reasoning in that detectives interpret clues that permit the course of events to be reconstructed (ALONSO, 1998). Abduction also resembles a doctor's process of reasoning when making inferences about the presence of illness based on symptoms²⁶⁾. Abduction has as its function to return rationality to the world when this rationality has been lost or questioned due to the presence of surprising or unexpected facts. But by doing so, it also contributes to our knowledge of the world by revealing and manifesting aspects of that world that were not previously taken into consideration. Surprise is not the consequence of defective or anomalous scientific practice, but is instead the basis of scientific discovery. Qualitative methodology provides the conditions for the unexpected to emerge in discourse insofar as it is an open and flexible methodology that encourages the manifestation of what is implicit and that which must emerge. The sociological interpretation of discourse, in that it is an application of abductive logic, therefore provides us with tools to deal with the unexpected in a scientific manner. [55]

In practice, the sociological analysis of discourse is conducted simultaneously on the three levels described above (textual, contextual and interpretive) in an on-going circular process between each of the levels until the research objective is achieved. In order to better explain this analytical process, in this section an example is presented as well as a summary of some aspects of the analysis of a text fragment²⁷⁾. Nonetheless, it should be noted that sociological analysis only focuses on text fragments at an advanced stage of the overall analysis. In a first instance, the sociologist undertakes a complete reading of the literal transcriptions and makes notes about the analysis with a view to understanding the meaning of the discourse as a whole. After this thorough and complete reading of the transcription, the text is broken down in a logical manner according to the researcher's understanding of the texts and contexts as well as his or her research objectives. [56]

The fragment of text presented in this section is taken from the transcription of a discussion group with manual workers who work in cooperatives in a town on the coast of Almeria, Spain²⁸⁾. Although the initial topic proposed was "immigration," the discussion progressively focused on questions of particular interest to the research aim. The discussion groups lasted between approximately 70-100 minutes. The group from which the fragment is drawn lasted 78 minutes. The selected fragment comes from a central part of the encounter, which took place around 35 minutes after the discussion had begun.

W1: I ..., my beautician ..., she ..., Moorish women²⁹⁾ have eyebrows up to here, don't they? And moustaches down to here. But they have to shave their twats. (laughter)

W1: Have it shaved. See, they really are racist, huh? See, they really are racist. They can't pluck their eyebrows but they have to shave their twats. And you have to go and eat it with hairs!

Four women of the group take part in this fragment. One of them (W1) plays a central role, commenting on the supposed aesthetic and hygienic customs of Moroccan women, who they refer to as las moras. The others act as a chorus, two of whom agree with the comments made by the first (W2 and W3) and a third (W4) who shows surprise at the customs. Yet the comment made by the last woman supports the argument of the first, since by expressing her surprise she is reaffirming the odd nature of the customs that the women are speaking about. One of the most significant features of the discussion group from which this fragment has been drawn is the spontaneous and practically immediate unanimity of the participants during the encounter, thus indicating a monolithic discourse without fissures in which dissent or disagreement do not occur and widespread consensus is reached among the participants on the proposed topic (immigration and immigrants). The general tone of the conversation is a relaxed one in which fun is made of immigrants of Moroccan origin. The laughter at the end of this fragment indicates that the participants find the customs of Moorish women strange and funny, as well indicating the general agreement among the group regarding this comment. [58]

The fragment was coded as a criticism of Moroccan immigrants, with a pejorative or insulting overtone. However, a close reading of the fragment leads us to code a twofold criticism: on the one hand the women criticize the supposed chauvinism of Moorish men, while on the other they also implicitly criticize the supposed submissiveness of Moorish women. This argument stems from the fact that the participants interpret the hygienic and aesthetic customs of Moorish women as being imposed by their husbands or partners. They are, therefore, derogatory comments directed specifically at Muslim immigrants. [59]

The criticism of immigrants can be interpreted as a sign of rejection (racism and/or xenophobia). This rejection is particularly evident in the pejorative manner in which the members of the group unanimously refer to the immigrants. Moreover, the derogatory comments expressed by this group can be interpreted as a sign of rejection not only by the social sphere where this group was formed (a small town on the coast of Almeria), but also by the social group to which they belong (low-skilled, middle-aged workers with a low educational level). [60]

In the majority of the discussion groups that were conducted, participants directed much of their criticism towards immigrants. Only two out of all the discussion groups made no pejorative comments, although veiled criticisms were made in all of them to a lesser or greater degree. In the group from which the above fragment was drawn, 46 derogatory comments were made about immigrants, of which 35 referred specifically to Muslim immigrants. This greater frequency of pejorative references is due, in part, to the group dynamics proposed by the moderator, who progressively focused the topic of conversation towards Muslim immigrants. Consequently, 25 minutes after the discussion had begun, the conversation centered almost exclusively on Muslim immigrants. For that reason, this is one of the discussion groups in which a greater number of pejorative comments towards immigrants were made in general and towards Muslims in particular, although many negative comments were also made in two other groups. [61]

As we have said, the fragment was coded in a twofold manner as a critical-derogatory allusion towards Muslim immigrants regarding their aesthetic and hygienic customs: the supposed chauvinism of the men and the supposed submissiveness of the women. However, a detailed reading reveals the existence of a third criticism of Muslim immigrants, namely their supposed racism. Indeed, in her last intervention, W1 refers to the racism of Moorish men ("See, they really are racist"), in what appears to be a confusion: it seems that what she really meant to say was "sexist," since this would be the term that corresponds to her argument and what she actually says bears no relation to "racism." This confusion may have arisen due to a metonymic transfer of the meaning: "racism" and "sexism" belong to the same category of attitudes that are condemned or looked down upon socially. [62]

However, there are contextual elements that lead us to suspect that the confusion is not due to a matter of proximity, but rather to a conceptual stretching: in reality she wanted to say "racist," although this label does not correspond to her criticism of the aesthetic customs of Moorish women. There are contextual reasons for thinking that this metonymic displacement is not the result of confusion, but rather an argumental strategy insofar as the speaker is interested in demonstrating the racism of Muslim immigrants. [63]

Firstly, we must bear in mind the defensive attitude taken by the group regarding the topic proposed by the moderator. Initially, the group's discourse developed along very moderate lines, with only a few general criticisms being made about the growing number of immigrants coming to Spain in recent years. But as the discussion progressed, the discourse took a more radical turn with an increasingly larger number of derogatory or insulting comments, as the fragment shows. The defensive attitude shown by the participants is likely due to the way they interpreted the encounter and the proposed topic of discussion. In other words, they perceived the fact of being brought together to speak about immigration as an implicit accusation of racism. Indeed, the widely-reported racist outbreak that occurred in the year 2000 in a nearby town (an event that the participants explicitly refer to at the end of the discussion) is another factor that heightened their distrust or sensitivity towards the topic as the residents in the area feel that they have been blamed for the outbreak. [64]

Because the participants perceive that they are being accused of racism in a veiled manner, the group initially uses a strategy to conceal their opinions (by moderating them). However, as the discussion progresses, rejection of immigrants in general, and immigrants of Muslim origin in particular, becomes increasingly evident. Thus, the group's discourse converses with (or responds to) the dominant discourse in society condemning racism (intertextuality). Although this has not been explicitly proposed, it is implicitly associated to the purpose of the encounter and to the question posed by the moderator. [65]

Following this initial strategy of concealment, the discourse drifts towards a counter-argument. No longer do the participants attempt to negate their rejection of immigrants, but to explain or justify it. Thus the argument turns into one of inversion through exaggeration: the terms are inverted by exaggerating and generalizing supposed concrete cases. For example, the participants state that the Spanish are discriminated against compared to immigrants in terms of access to public resources. According to them, not only are immigrants not discriminated against, but they are granted special privileges in terms of public services. Another example of this type of argument that arose in the majority of the discussion groups is the affirmation that immigrants are more racist than ‘we are'. To a certain degree, this supposedly greater racism would excuse and justify ours. It is precisely this second argument that arises in response to the accusation of racism where references to the racism of Muslims is analyzed in terms of conceptual stretching. Given the participant's interest in demonstrating such racism and the difficulties involved in providing proof of it, her arguments become increasingly stronger. It is not that she meant to say "sexist" and out of confusion said racist. What she really meant to say was racist, even though there was no direct relationship between racism and her criticism: the meaning is metonymically displaced to make it coherent with the argument the participant wishes to make, even if this means that the argument is unfounded and largely incomprehensible. The fragment analyzed shows the difficulties the group encounters when attempting to defend the supposed racism of Muslim immigrants, but it also shows how, in spite of these difficulties, the group develops discourse strategies to sustain their argument. In short, the analyzed fragment contains three criticisms of Muslim immigrants: the supposed sexism of the men, the supposed submissiveness of the women and the supposed racism of both. [66]

The analysis presented here permits us to make three interpretive conjectures. These conjectures are not put forward so much as empirical evidence (to contrast with the hypothesis), but as indicators of underlying realities and social processes that explain the discourses studied (to formulate the hypothesis). These conjectures are revised and contrasted in the process of analysis, that is, in comparison with other textual fragments and as part of the full analysis of the transcriptions. Specifically, we highlight four interpretive conjectures of the analysis of this textual fragment:

Two questions were posed in the introduction. Firstly, beyond their apparent diversity, what elements do the different approaches to sociological discourse analysis share, and secondly, what differentiates this sociological approach from other analytical approaches? In this article, we argue that the elements these approaches have in common are precisely what differentiate them from other approaches, namely: a) their eclectic character, in that they are all structured around methods of a diverse origin and analytical nature and b) the type of interpretation they propose, that is, the linking of discourse with broader social realities. [68]

While diverse forms of textual and contextual analysis are a part of sociological discourse analysis, they are not in themselves sociological analyses since what distinguishes these from other approaches to discourse is the type of interpretation they propose. Sociologists frequently resort to different procedures for textual and contextual analysis. According to the approach we propose here, however, these are only intermediate stages or phases of analysis aimed at providing an interpretation that connects the analyzed discourses with the social context in which they have emerged and circulate. [69]

On occasion, exclusively textual or contextual analyses are presented as sociological discourse analyses. In this case, the sociological interpretation of discourse is implicitly derived in a direct or immediate manner from the textual or contextual analysis presented. The formulation of this type of interpretation does not differ substantially from the sociological analysis proposed here. The problem lies in the fact that when the jump to interpretation is made in an implicit way, criticism is impeded or, is at least made difficult. Moreover, sociological discourse analysis is based on a combination of both textual and contextual techniques and procedures of analysis in order to improve the reliability of the interpretations. For this reason, sociological interpretations of discourse that are formulated on the basis of a single procedure of analysis are risky at best³⁰⁾. [70]

The validity of sociological discourse interpretations derives from a criterion of intersubjectivity: once the materials of analysis have been examined, the interpretation should be considered valid by anyone who evaluates it in a critical manner. Only if this requisite of intersubjectivity is fulfilled will discourse analysis achieve its objective, that is, to further our understanding and knowledge of social phenomena. It is for this reason that interpretation must be formulated in an explicit manner. For this same reason it is also important to explain the logic on which these interpretations are based since they are uncommon, little known and often recognized as problematic. [71]

This article explores the theoretical criteria and presuppositions of discourse analysis based on the author's own research experience at the Qualitative Unit of the Institute for Advanced Social Studies (IESA/CSIC). This paper has been made possible thanks to the work conducted by the author in conjunction with his colleagues Ángel RAMÍREZ, José Antonio CERRILLO and Luís RODRÍGUEZ-MORCILLO. The author is also grateful to Fernando AGUIAR, whose valuable comments and criticisms have contributed greatly to improving the paper.

The author is grateful to Martha GAUSTAD for her translation from Spanish into English of this paper.

1) This diversity of what is understood by sociological discourse analysis is reflected in papers published by FQS in recent years. For example, articles and book reviews about qualitative content analysis (SPANNAGEL, GLÄSER-ZIKUDA & SCHROEDER, 2005; MAYRING, 2000; FAUX, 2000), automatic discourse analysis (HELSLOOT & HAK, 2007), grounded theory (MERLINO & MARTÍNEZ, 2006; KELLE, 2005), Foucaultian discourse analysis (DÍAZ-BONE, 2005; ROMÁN, 2007; AMIGOT, 2007), critical discourse analysis (KENDAL, 2007; DIRKS, 2006), conversational analysis (ASHMORE & REED, 2000; HAVE, 2006; BERKENBUSCH, 2009), or interpretative visual analysis (SCHNETTLER & RAAB, 2008) have all featured in FQS. Furthermore, FQS has even published an article dedicated to positioning analysis; a sort of middle-ground position between critical discourse analysis and conversation analysis (KOROBOV, 2001). According to the approach defended here, far from being a problem, this diversity reveals the wealth and possibilities of discourse analysis for sociology. <back>

2) It is not difficult to find examples of sociological discourse analyses which lack in rigor, are insufficiently grounded or even clearly inaccurate. In this sense, an interesting review and critique of these non-rigorous analytical practices can be found in ANTAKI, BILLIG, EDWARDS and POTTER (2003). However, it would seem unjustified to extend this criticism to the majority of discourse analyses conducted by sociologists or even a significant portion of them. <back>

3) For example, Roland BARTHES' analyses on eating as a discourse are classics. For a critical review of these analyses see ALONSO (2005). <back>

4) This tiered approach to sociological discourse analysis is similar to that proposed by ALONSO (1998). Nonetheless, there are important differences between the two approaches. On the one hand, the first two levels proposed by ALONSO—the informational and the structural level—are englobed in our approach at an initial textual level. On the other hand, we refer to a second contextual level which ALONSO includes in his third level, the strictly sociological or interpretive level. In spite of this, the two approaches have more similarities than differences in two ways: both consider the informational and the structural levels of analysis to be insufficient or merely complementary to the sociological analysis and both consider sociological interpretation as a distinctive or specific element of this sociological analysis. <back>

5) For a more in-depth review of the methods of content analysis see NAVARRO and DÍAZ (1994) and ANDREU (2002). <back>

6) The main principles and procedures of Grounded Theory can be found in GLASER and STRAUSS (1967). GLASER (1992) provides a review and updated version of grounded theory. See also VALLES (2000) and ANDREU, GARCÍA-NIETO and PÉREZ CORBACHO (2007). <back>

7) Although the terms textual and literal may, on occasion, be used interchangeably, the latent meanings of discourse reveal a fundamental difference: texts can, and in fact usually do, include meanings which are different or even contrary to their literal form. <back>

8) The attention paid to the textual structure means that greater consideration is given to the context in which the textual "fragments" are found. When speaking of context in this sense, we refer to the relationships between the different elements of the text when considering the text as a whole. This is the intratextual context (utterance level), while other "contexts" are considered at other levels of analysis. Here we consider the intertextual and the situational context at a second contextual level (enunciation level), while the broader social context is considered at a third, strictly sociological level of analysis (level of sociological interpretation). <back>

9) For a review of the applications of network theory to textual discourse analysis see LOZARES, VERD, MARTÍ and LÓPEZ (2003). Theory of Argumentation is of special interest to sociological analysis (ANSCOMBRE & DUCROT, 1994). Practical applications of this theory can be found in MARTÍ (2006). <back>

10) Various methods or procedures are used to conduct structural semiotic analysis, the majority of which are drawn from linguistics, in particular, structural linguistics. A specially interesting overview of techniques for semiotics analysis can be found in ABRIL (1994). <back>

11) For a review of the main criticisms of structural semiotic analysis see ALONSO and CALLEJO (1999). <back>

12) We can also speak of an intratextual context as the largest unit of the text of which an element of that text forms part. Nonetheless, textual analysis accounts for this context. For this reason we have referred to it above when discussing the contribution of qualitative approaches to content analysis in relation to the need to contextualize the analysis. <back>

13) The analysis of discourse positions also suffers from certain limitations that chiefly affect the analysis of collective or group discourses. The analysis of the discourse positions held by subjects within a discussion group cannot be understood as the atomization of discourses to facilitate analysis. For example, we cannot speak of a female discourse within mixed group discussions. To capture female discourse on a given topic we must have discussion groups that are differentiated by sex in order to obtain different discourses and compare them. To put it another way, the discourse position adopted by women on a given topic in a mixed discussion group differs from the female discourse on this same topic. <back>

14) Interviews or discussion groups can be understood as situations of communicative exchange that are regulated or defined by the social researcher. <back>

15) If this is true for formal situations established in the framework of social research, it is even more so for informal situations in which subjects produce discourse on an everyday basis. <back>

16) The dialogic nature of discourse situations becomes clearer and stronger when communicative feedback is possible among the subjects involved. This is the case, for example, of face-to-face conversations. Nonetheless, it is a characteristic of all the situations in which discourse is produced, including monologues, opinion articles or in the discourse of advertising; all of which are produced under situations in which the discourse would seem to respond solely to the discretional will of the sender. The dialogic nature of the situation is common to all discursive production with the exception perhaps of individuals with serious psychiatric problems. <back>

17) An interesting and thorough review of the procedures and principles of conversation analysis can be found in GALLARDO-PAÚLS (1996). Also see ANTAKI and DÍAZ (2003), LEVINSON (2004) and TUSÓN (1997). <back>

18) In the following section we will return to the ideological interpretation of discourse that is characteristic of Critical Discourse Analysis (CDA), of which FAIRCLOUGH is a principal founder. <back>

19) For an interesting reflection upon the relationship between knowledge, ideology and discourse see VAN DIJK (2005). <back>

20) In the strict sense, induction involves verifying the predictions that have been derived deductively from the theory. However, this induction-cum-generalization—"we induce when we generalize from a number of cases that something is true, inferring that the same is true for all the class" (SANTAELLA, 1998, p.6)—does not differ substantially from the verification of a theoretical prediction. Indeed, generalizations are formulated based on our prior theoretical conceptions, which direct our interest towards specific information in the discourse. In this sense, they are little more than the provisional acceptance of a theoretical premise provided that there is no empirical evidence to the contrary. <back>

21) This characteristic of social discourse samples and, in general, of qualitative social research has been dealt with in a very clarifying manner by SACKS (2000) through an analogy with the study of grammar: by analyzing how an individual uses language, we can know in a very broad and certain manner the grammatical rules of that language. It is not so much a question of accumulating empirical evidence, but of detecting individuals who due to their particular use of language can present variations on the grammatical rules that they use in their speech. <back>

22) We can therefore speak of two types of unexpected or unforeseeable results in sociological discourse analysis: those that broaden our knowledge of the social reality and those that modify our view or concept of it, that is, our theories about what is real. <back>

23) In the seventh and last of his lectures given at Harvard on 14 May 1903, PEIRCE formulated a process for abductive inference in the following terms "The form of inference, therefore, is this: The surprising fact, C, is observed; But if A were true, C would be a matter of course, Hence there is reason to suspect that A is true (PEIRCE, 1903, p.189). <back>

"It must be remembered that abduction, although it is very little hampered by logical rules, nevertheless is logical inference, asserting its conclusion only problematically or conjecturally, it is true, but nevertheless having a perfectly definite logical form. (...) Only in deduction that there is no difference between a validargument and a strongone. (...) An argument is none the less logical for being weak, provided it does not pretend to a strength that it does not possess (PEIRCE, 1903, p.192).

The suspicions aroused by sociological discourse interpretation can be understood as a consequence of this jump to abductive interpretation: sociological interpretation says "something more" than, or goes "beyond" the contents of the discourse, but for this reason it does so in a weak manner as a hypothesis or conjecture. <back>

25) However, induction in sociological discourse analysis also responds more to a logic of discovery than a logic of verification. <back>

26) The type of reasoning in which facts are interpreted as symptoms has recently become very popular thanks to the television series House, which deals with cases occurring in a hospital unit specialized in difficult diagnoses. These cases contain all the ingredients of abduction: a surprising fact in the form of an illness that is difficult to diagnose, the "interpretive" leap towards diagnoses based on the patients' symptoms and the verification or empirical testing of diagnoses formulated in an abductive or hypothetical manner through a variety of medical tests. This is a good example of abductive thought in the form of tentative reasoning which, operating through a series of attempts, orients and requires the continual empirical verification and progressive refinement of the formulated hypotheses until a plausible explanation for the surprising fact is found. <back>

27) For readers interested in more in-depth examples of sociological discourse analysis, see the publications of CONDE (1999, 2002, 2007). <back>

28) This is one of eleven groups that participated in a study conducted by the IESA-CSIC under commission of the State Secretariat for Immigration and Emigration on perceptions and attitudes about Islam and Muslims in Spain. The fieldwork was conducted from December 2007 to January 2008, while the analysis and reports were completed from February to May of this same year. <back>

29) In the original text in Spanish, the participants refer to Moroccan men and women as los moros and las moras, respectively. This is a derogatory term used in Spanish to refer to Muslims of North African origin. Here we will use the term "Moorish men" or "Moorish women" given that it carries similar negative connotations. <back>

30) In this regard, the case of frame analysis is illustrative since it is very common to derive ideological interpretations from situational communicative frameworks, that is, to derive the presence of ideologies held by the subjects involved. There is nothing wrong with this type of analysis, provided that the process of interpretation is clearly explained and justified. <back>

Abril, Gonzalo (1994). Análisis semiótico del discurso. In Juan Manuel Delgado & Juan Gutiérrez (Eds.), Métodos y técnicas cualitativas de investigación en ciencias sociales (pp.427-464). Madrid: Síntesis.

Alonso, Luís Enrique (1988). Entre el pragmatismo y el pansemiologismo. Notas sobre los usos (y abusos) del enfoque cualitativo en sociología. Revista Española de Investigaciones Sociológicas, 43, 157-170.

Alonso, Luís Enrique (1998). La mirada cualitativa en sociología. Barcelona: Fundamentos.

Alonso, Luís Enrique (2002). Los mercados lingüísticos o el muy particular análisis sociológico de los discursos de Pierre Bourdieu. Revista de Estudios de Sociolingüística, 3(1), 111-132.

Alonso, Luís Enrique (2005). Mitologías alimentarias cotidianas: una relectura de Roland Barthes. Revista Internacional de Sociología, 40, 79-107

Alonso, Luís Enrique & Callejo, Javier (1999). El análisis del discurso: del postmodernismo a las razones prácticas. Revista Española de Investigaciones Sociológicas, 88, 37-74.

Amigot Leache, Patricia (2007). Más allá del discurso: Análisis genealógico de un proceso de transformación intersubjetiva de género. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 9, http://nbn-resolving.de/urn:nbn:de:0114-fqs070295 [Date of access: 29.03.09].

Andreu, Jaime (2002). Las técnicas de análisis de contenido: una revisión actualizada. Sevilla: Fundación Centro de Estudios Andaluces.

Andreu, Jaime; García Nieto, Antonio & Pérez Corbacho, Ana María (2007). Evolución de la teoría fundamentada como técnica de análisis cualitativo. Madrid: CIS.

Anscombre, Jean-Claude & Ducrot, Oswald (1994). La argumentación en la lengua. Madrid: Gredos.

Antaki, Charles & Díaz, Félix (2003). El análisis de la conversación y el estudio de la interacción social. In Lupicinio Íñiguez (Ed.), Análisis del discurso: manual para las ciencias sociales (pp.125-140). Barcelona: UOC.

Antaki, Charles; Billig, Michael; Edwards, Derek & Potter, Jonathan (2003). El análisis de discurso implica analizar: crítica de seis atajos analíticos. Athenea Digital, 3, http://antalya.uab.es/athenea/num3/antaki.pdf [Date of access: 17.01.09].

Ashmore, Malcolm & Reed, Darren (2000). Innocence and nostalgia in conversation analysis: The dynamic relations of tape and transcript. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 1(3), Art. 3, http://nbn-resolving.de/urn:nbn:de:0114-fqs000335 [Date of access: 29.03.09].

Berelson, Bernard (1952). Content analysis in communication research. Glencoe: Free Press.

Berkenbusch, Gabriele (2009). El análisis conversacional como método para la investigación del aprendizaje intercultural – resumen de un proyecto extracurricular combinando aprendizaje e investigación. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 10(1), Art. 33, http://nbn-resolving.de/urn:nbn:de:0114-fqs0901335 [Date of access: 29.03.09].

Conde, Fernando (1999). Los hijos de la desregulación. Jóvenes, usos y abusos en los consumos de drogas. Madrid: CREFAT-Cruz Roja.

Conde, Fernando (2002). La mirada de los padres: Crisis y transformación de los modelos de educación de la juventud. Madrid: CREFAT-Cruz Roja.

Conde, Fernando (2007). Metropolización, territorio y vivienda en Andalucía. Culturas e identidades urbanas. Sevilla: Consejería de Obras Públicas y Transportes.

Debrock, Guy (1998). El ingenioso enigma de la abducción. Analogía: Revista de Filosofía, Investigación y Difusión, XII(1), http://www.unav.es/gep/AN/Debrock.html [Date of access: 23.01.09].

Díaz-Bone, Rainer; Bührmann, Andrea D.; Gutiérrez Rodríguez, Encarnación; Schneider, Werner; Kendall, Gavin & Tirado, Francisco (2007). El campo del análisis del discurso Foucaultiano. Características, desarrollos y perspectivas. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 30, http://nbn-resolving.de/urn:nbn:de:0114-fqs0702305 [Date of access: 29.03.09].

Dirks, Una (2006). How critical discourse analysis faces the challenge of interpretive explanations from a micro- and macro-theoretical perspective. Review essay: Gilbert Weiss & Ruth Wodak (Eds.) (2003). Critical discourse analysis. Theory and interdisciplinarity. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 7(2), Art. 26, http://nbn-resolving.de/urn:nbn:de:0114-fqs0602261 [Date of access: 29.03.09].

Fairclough, Norman (1995). Critical discourse analysis. Harlow: Longman Group UK Ltd.

Faux, Robert (2000). A description of the uses of content analyses and interviews in educational/ psychological research. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 1(1), Art. 26, http://nbn-resolving.de/urn:nbn:de:0114-fqs0001265 [Date of access: 29.03.09].

Gallardo Pauls, Beatriz (1996). Análisis conversacional y pragmática del receptor. Valencia: Episteme.

Glaser, Barney G. (1992). Basics of grounded theory analysis. Mill Valley, CA: Sociology Press.

Glaser, Barney G. & Strauss, Anselm L. (1967). The discovery of grounded theory. Chicago: Aldine.

Goffman, Erving (1986). Frame analysis: An essay on the organization of experience. Boston: Northeastern University Press.

Have, Paul ten (2005). Conversation analysis versus other approaches to discourse. Review essay: Robin Wooffitt (2005). Conversation analysis and discourse analysis: A comparative and critical introduction. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 7(2), Art. 3, http://nbn-resolving.de/urn:nbn:de:0114-fqs060239 [Date of access: 29.03.09].

Helsloot, Niels & Hak, Tony (2007). Pêcheux's contribution to discourse analysis. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 1, http://nbn-resolving.de/urn:nbn:de:0114-fqs070218 [Date of access: 29.03.09].

Herrera, Manuel & Soriano, Rosa María (2004). La teoría de la acción social en Erving Goffman. Papers, Revista de Sociología, 73, 59-79.

Hoffmann, Michael (1998). ¿ Hay una "lógica" de la abducción? Analogía: Revista de Filosofía, Investigación y Difusión, XII(1), http://www.unav.es/gep/AN/Hoffmann.html [Date of access: 23.01.09].

Ibáñez, Jesús (1979). Más allá de la Sociología. El grupo de discusión. Técnica y crítica. Madrid: Siglo XXI.

Ibañez, Jesús (1985). Análisis sociológico de textos o discursos. Revista Internacional de Sociología, 43(1), 119-160.

Kelle, Udo (2005). "Emergence" vs. "forcing" of empirical data? A crucial problem of "grounded theory" reconsidered. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(2), Art. 27, http://nbn-resolving.de/urn:nbn:de:0114-fqs0502275 [Date of access: 29.03.09].

Kendall, Gavin (2007). What is critical discourse analysis? Ruth Wodak in conversation with Gavin Kendall. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 29, http://nbn-resolving.de/urn:nbn:de:0114-fqs0702297 [Date of access: 29.03.09].

Korobov, Neill (2001). Reconciling theory with method: From conversation analysis and critical discourse analysis to positioning analysis. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 2(3), Art. 11, http://nbn-resolving.de/urn:nbn:de:0114-fqs0103119 [Date of access: 29.03.09].

Levinson, Stephen C. (2004). Significados presumibles: la teoría de la implicatura conversacional generalizada. Madrid: Gredos.

Lozano, Jorge & Peña-Marín, Cristina (1988). Discurso. In Roman Reyes (Ed.), Terminología científico social. Aproximación crítica (pp.294-297). Barcelona: Anthropos.

Lozares, Carlos; Verd, Joan Miquel; Martí, Joel & López, Pedro (2003). Relaciones, redes y discurso: revisión y propuestas en torno al análisis reticular de datos textuales. Revista Española de Investigaciones Sociológicas, 101, 175-200.

Martí, Joel (2006). Representación de estructuras argumentativas mediante el análisis de redes sociales. REDES: Revista Hispana para el Análisis de Redes Sociales, 10, http://revista-redes.rediris.es/pdf-vol10/vol10_4.pdf [Date of access: 12.02.09].

Merlino, Aldo & Martínez, Alejandra (2006). Integración de métodos cualitativos y cuantitativos: Construyendo e interpretando clusters a partir de la teoría fundamentada y el análisis del discurso. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(1), Art. 21, http://nbn-resolving.de/urn:nbn:de:0114-fqs0701219 [Date of access: 29.03.09].

Navarro, Pablo & Díaz, Capitolina (1994). Análisis de contenido. In Juan Manuel Delgado & Juan Gutiérrez (Eds), Métodos y técnicas cualitativas de investigación en ciencias sociales (pp.177-224). Madrid: Síntesis.

Peirce, Charles S. (1901). Sobre la lógica de la extracción de la historia a partir de documentos antiguos, especialmente de testimonios, http://www.unav.es/gep/LogicOfDrawingHistory.pdf [Date of access: 10.03.09]

Popper, Karl R. (1965). La lógica del descubrimiento científico. Madrid: Tecnos.

Potter, Jonathan & Wetherell, Margaret (1987). Discourse and social psychology. Beyond attitudes and behaviour. London: Sage.

Román Brugnoli, José Antonio (2007). Lo que las metáforas obran furtivamente: discurso y sujeto. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 12, http://nbn-resolving.de/urn:nbn:de:0114-fqs0702122 [Date of access: 29.03.09].

Sacks, Harvey (2000). Sobre muestreo y subjetividad. In Félix Díaz (Ed.),Sociologías de la situación (pp.85-94). Madrid: La Piqueta.

Santaella, Lucía (1998). La evolución de los tres tipos de argumento: abducción, inducción y deducción, http://www.unav.es/gep/AN/Santaella.html [Date of access: 25.01.09]

Schnettler, Bernt & Raab, Jürgen (2008). Interpretative visual analysis. Developments, state of the art and pending problems. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 9(3), Art. 31, http://nbn-resolving.de/urn:nbn:de:0114-fqs0803314 [Date of access: 29.03.09].

Schutz, Alfred (1974a). El problema de la realidad social. Buenos Aires: Amorrortu.

Spannagel, Christian; Gläser-Zikuda, Michaela & Schroeder, Ulrik (2005). Application of qualitative content analysis in user-program interaction research. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(2), Art. 29, http://nbn-resolving.de/urn:nbn:de:0114-fqs0502295, [Date of access: 29.03.09].

Valles, Miguel S. (2000). La grounded theory y el análisis cualitativo asistido por ordenador. In Manuel García Ferrando, Jesús Ibáñez & Francisco Alvira (Eds.), El análisis de la realidad social (pp.575-603). Madrid: Alianza.

Van Dijk, Teun A. (1999). El análisis crítico del discurso. Anthropos: Huellas del Conocimiento, 186, 23-36.

Van Dijk, Teun A. (2005). Discurso, conocimiento e ideología. Reformulación de viejas cuestiones y propuesta de algunas soluciones nuevas. Cuadernos de Información y Comunicación, 10, 285-318.

Jorge RUIZ RUIZ is B.A. Degree in Sociology from the Complutense University of Madrid (UCM) and was awarded the Complutense Prize for Research in 1992. His professional career got off to a start in 1993 with the creation of his own company, Caleidoscopia Investigación Social S.L. He later collaborated with several firms and institutions in the field of social and market research. Since 2002, Jorge Ruiz has worked at the Institute for Advanced Social Studies (IESA-CSIC) as a technician specialized in qualitative research methods. His research interests have centered chiefly on the reality and problems of youth and the use and development of qualitative methodology.

Instituto de Estudios Sociales Avanzados (IESA/CSIC)
Campo Santo de los Mártires, 7
14004 Córdoba, Spain

Ruiz Ruiz, Jorge (2009). Sociological Discourse Analysis: Methods and Logic [71 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 10(2), Art. 26, http://nbn-resolving.de/urn:nbn:de:0114-fqs0902263.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License

Type of analysis	View of discourse	Level of analysis	Methods or procedures of analysis	Objectives
Textual analysis	As object	Utterance level	Content analysis Semiotic analysis (structural and formal)	Characterization of discourse
Contextual analysis	As singular event	Enunciation level	frame analysis, analysis of discourse positions, conversation analysis, intertextual analysis	Understanding discourse
Sociological interpretation	As information, ideology and social product	Social level	Inductive inference, abductive inference	(Sociological) explanation of discourse