CC BY
202
TEXTUAL ANALYSIS WITHIN SOCIOLOGICAL RESEARCH: TERMINOLOGICAL RATHER THAN METHODOLOGICAL DIFFICULTIES?
Irina V. Trotsuk
Sociology Chair Peoples' Friendship University of Russia
Miklukho-Maklaya str., 10/2, Moscow, Russia, 117198
The article aims at providing some clear distinctions between different types of textual analysis that have different but sometimes interconnected theoretical and methodological grounds. Terminological confusions and the lack of clear-cut instructions for the empirical study of textual data within sociological research impede such research.
Key words: textual analysis, narrative analysis, discourse analysis, conversational analysis, linguistic analysis, content analysis, coding.
Whether or not you are a textual analysis expert you'll easily name a lot of types of such if asked to do so. Everybody seems to know at least what discourse analysis and content analysis are in the contemporary world where the lay and the competent knowledge have drawn closer due to the web-search resources. The more aware researcher would probably name a lot more types of analytical work with differently formalized textual data, such as narrative analysis, biographical analysis, document analysis, conversational analysis, linguistic analysis, sequence analysis and so on besides word combinations with term 'coding' (textual, open, selective, thematic etc.). Most scientific articles use all these terms as quite obvious and clear rarely identifying any slight or serious differences between them if that seems important for choosing the specific research tool [1].
In most disciplinary fields such position may be appropriate, but that's quite inexcusable within sociological frames due to the serious borders as well as intersections between the theoretical and empirical levels here. If we are to conduct an empirical sociological research we have to find definite unambiguous empirical indicators to all semantic constructions that identify the object, the subject and the categories of the textual analysis that may go without any saying within other disciplinary frames (for instance, 'narrative' and 'discourse' in the philosophical tradition, 'ideological discourse' in political studies etc.). So let's try to look at the textual analysis typology from the point of view that would help us understand what empirical procedures are expected to be used or supposed by the most 'popular' and frequently mentioned in scientific literature types of textual analysis (discourse, narrative and conversational, not the text/textual analysis for the latter seems empty and infinitely full at the same time). It is not the primary intention of this article to enter into the specificity of the textual approaches debate, but rather to contribute to the wider discussion about the criteria for conducting empirical textual analysis and to set one possible perspective alongside those of other analysts.
The term narrative analysis stresses the object of sociological study in the first place — that is the narrative or narrative form of empirical sociological evidence [9. P. 517] (the same is the situation with the term biographical analysis that focuses on the study of so called socio-biographical data) which can be 'dissected' with different tools and procedures. The classic definition of narrative was given by W. Labov: "method of recapitulating past experience by matching a verbal sequence of clauses to the sequence of events which (it is inferred) actually occurred" [11. P. 359—360]. This definition has survived, more or less intact, through the years and through a huge number of different interpreters that provided pretty convincing evidence of almost any "perceived sequence of nonrandomly connected events" [16. P. 7] (verbal and visual, fiction and non-fiction, personal and group stories) to be narrative.
Most of the prominent founders of narrative analysis proposed to describe and classify either narratives themselves or their basic characters as well as other important elements of the story. According to A.J. Greimas, all narratives have six basic ac-tants, working in sets of three interrelated pairs — sender/receiver, helper/opponent, subject/object. V. Propp identified an invariant pattern of 31 functions behind the large variety of Russian folktales. W. Labov found a six-part macrostructure in narratives of New York Harlem African-Americans — abstract, orientation, complicating action, evaluation, result, resolution and coda [11. P. 362—370]. We can continue the list. The more important is to admit the following facts: firstly, the structural approach is the most popular and widely used, especially in the western research tradition; secondly, we identify the structural parts of narrative through the linguistic patterns pointing to them; finally, there are no (for it's impossible) definite ways to fulfilling such a task at the empirical level of analysis, so we can receive absolutely different interpretations of the same narratives by different researchers if there are no strict limits to analytical work similar to the classical content analysis coding scheme.
Of course, dealing with narrative data we may address its linguistic properties (as our clues to interpretation), but usually researchers stress the object of the study not the specifics of analytical approach (i.e. the corresponding linguistic markers of dynamic motifs, cardinal functions and other top points of the story told to the researcher by informants with quite different levels of narrative competence), so we don't automatically shift out terminology to the linguistic analysis in such cases. Let's say a few words about it as well.
Linguistic analysis and the study of literature may seem to share the object of research, but the former is wider for we are interested in all types of texts, not only the fiction, and it considers all linguistic elements of the test as pointing to the main idea not vice versa — analytical work with the meaning to explain the chosen aesthetics, genre and composition design. Consequently within the literature study the researcher must focus on the author position and try to explain it appealing to the wide philosophical, historical and social tradition, whereas the linguist concentrates on this very specific textual data, trying to foresee the readers' reactions [32. P. 4]. That's why when conducting linguistic analysis we are to use such terms as 'inventions', 'dispositions', 'elocutions' and so on, consider the social and historical context of the text creation, its pragmatic functions and connections of general style/genre/language rules and au-
thor individual creativity. Sometimes the proposed complicated and diverse terminology in the sociological research is replaced by more 'empirically grasped' — we try to identify the factual information (the events reported and the main themes), the conceptual information (the author's interpretation of relationships and causal chains that can be seen only through the text as a whole), the subtext information (that is the result of linguistic units ability to produce associations and connotations); the objective (the author's perception of reality) and the subjective (how author evaluates his own words) modalities of the text [32. P. 27—35].
There are different empirical approaches to conducting linguistic analysis, and they can be summed in the following few types [25. P. 16—32]: stylistic experiments (rewriting the text using synonyms, other ways of organizing the data etc.); semantic method, i.e. the evaluation of departures from the language rules, reasons of multiple meaning and reiteration appearance etc.; comparative stylistics method — looking for differences and resemblances in describing the same content in a number of texts; the variety of all possible techniques of quantitative textual analysis (where we believe to discover qualitative peculiarity of the text through its quantitative characteristics).
It's quite obvious that the second and the third approaches look very much like discourse analysis. Linguistic and discourse studies appear to become almost indistinguishable when they both look for the so called 'register' — the communicative type of the text that depends on the spatio-temporal position of the author and his attitudes towards what he says. Usually we identify three basic registers: reproductive/figurative (the narrator speaks about the things directly observed and uses phrases with 'I see/hear/ smell/sense...'); informative (the storyteller reproduces information which happens to be the result of repeated observations and logic operations and uses phrases starting with 'I know/see...'); the generitive (the author generalizes one's experience contex-tualizing it with social models, historical trends and general maxims) [22. P. 28—29]. And that's in addition to the indisputable fact that within narrative and discourse analyses we pay attention to the linguistic markers that help to identify our informants as members of some speech and ideological (for instance, the language of power, the language of hate etc.) communities typically found among the specific ranks of the society.
The discourse analysis is a multidisciplinary research field, an umbrella name of variety of analytical approaches to textual data that differ by their research priorities. These may be symbolic exchanges in the communication, memory and understanding basic structures, the language of science, the social influence of the institution under study and so on. The platform that unites different, sometimes unconceivable together theoretical and methodological approaches is the belief that the social reality is constructed linguistically, i.e. for the purposes of sociological research we have to study it through the 'social texts' that represent not the reality itself but the way people see and describe it. It supposes that the discourse analysis considers any text only within the intertextual (social, cultural, political, group, individual etc.) conditions of its creation and existence [27].
It's impossible to form any sensible discourse analysis typology, but we can confidently name two main classical and traditional branches of discourse analysis — conditionally the 'sociological' and the 'linguistic' ones. The former was developed by M. Foucault who considered discourse as an ideologically determined way to talk and
pointed to the four measures of every discourse: objects, modality, concepts and thematic unity. According to Foucault, the discourse analysis intends to identify the possibility of 'talk' about specific objects, the type of phrases modality (who, where and what can say about these objects), the chance for the specific concepts to appear in the organized talk and the strategies for choosing themes and theories [28. P. 40]. So the discourse can control and subordinate owing to the outer (exclusion practices — bans, madness evaluation and the will to truth) and inner (classification and sorting out) procedures [29. P. 52, 65].
T. van Dijk developed 'linguistic' version of discourse analysis on the assumption that even the observed facts are inevitably and unconsciously described according to the specific personal picture of reality, that's why discourse can be considered the structuring principle of every communicative interaction [20]. The latter is determined by the situation model (combination of time, place, context, actions, participants, causes, aims, consequences, categories of appraisal) that may significantly vary letting a person understand quite different texts. So the discourse analysis supposes finding in the text the situational model elements and their combinations a priori set by the researcher to summarize and identify the ideological position of the author in some relevant categories (such as 'superstructure', 'macrostructure' etc.) [20. P. 45—60].
The 'traditional' discourse analysis suggested by Foucault has been severely criticized for 'radical ontological constructionism' (dropping out nonsemantic features of reality), 'nominalistic forms of conceptualization and explanation', 'underlying determinism' (discourses appear to be autonomous and independent from peoples' will), 'localism' (the structural stability of power hierarchies is underestimated), 'reduction of ideologies study to the discourse analysis' (without paying attention to the wide social, political, historical context that institutionalize discourse as a recourse of significant social meanings and systems of control) [13. P. 525—527]. The more 'realistic' version of discourse analysis is a proposed alternative: discourse is considered to be mediator between social actions and constitutive features of reality defining actors' positions in the matrix of social and linguistic rules and relationships. So the actors can achieve reinterpretation of the dominant discourses although all their strategic and tactical initiatives are built into the existing system of material, social and discursive practices that can't be ignored.
Another famous version of discourse analysis based on the term 'myth' was developed by R. Barthes who defines myth as "a word, communicative system, some message, form or way of constituting meaning that is historically framed, is subject to the conditions of applications and full of social content" [19. P. 234] (for example, a simple tree, a pure material object, may become a symbol of sad or happy memories and associations, i.e. filled with social meaning). Here we see that discourse and narrative analyses get very close in such interpretation of the former that causes some terminological confusion reinforced by the Barthes' suggestion that the semiology must be responsible for reading and decoding myths (as narratives and discourses — no principle differences between them are stated and the matter of such is out of question), considering linguistic and metalinguistic features of the text using terms 'targeting', 'functions', 'intentions', 'imperative response', 'fragmented meaning' etc.
Moreover, the terminological indistinguishability of discourse and narrative analyses becomes more obvious when the latter is viewed as a 'qualitative approach' (the same as narrative, biographical and interpretative analyses within sociological research) [6. P. 1140] to identify discursive frames determining our everyday perception of reality and ways to talk about it. We focus on the situational construction of communication and immediately the question arises what is the difference then between discourse and conversational analyses, although many researchers refuse to see and admit it. So it's quite evident that discourse analysis is not a uniform category of a single methodological orientation and loosely overlaps with conversation analysis and narratol-
ogy [8].
The narrative and discourse analyses are obviously theoretically oriented (of course in most cases we can claim only to build a mini-theory of the case under consideration or the 'grounded theory'), whereas conversational analysis is empirically oriented. That's why we don't experience serious terminological and interpretational difficulties here, although the researcher's personal preferences (either scientific or everyday) always interfere and affect the study results: "our ability to understand and fully grasp the meaning of the text.. is inextricably linked to a wealth of background knowledge that readers consciously or unconsciously bring to the text in the construction of meaning" [9. P. 545].
The conversational analysis is pretty clear in whatever considers the correlation of its methodological bases and empirical procedures. Another significant distinction of the conversational analysis is its focus on the extralinguistic rather than linguistic characteristics: for instance, for the purposes of narrative and discourse analyses we rarely if ever indicate the pauses in interviews or other types of data transcripts — conversation analysts, on the contrary, are firmly convinced that silences may speak louder than words [7. P. 107—114], for example, on gender differences or the fatality of the patient condition in interactions between doctors, patients and their families [12. P. 118]. The conversationalists start from the empirical evidence without the premeditated hypotheses, treat the smallest talk/text details as an important analytical resource rather than undesirable 'noises', proceed from the assumption that the social order of speech is 'visible' not only for the researchers but for people constructing it as well because they are more competent in their daily routine than the supposedly expert analysts (that's the basic imperative of ethnomethodology) [23. P. 37]. The latter trait of conversational analysis means that the gender study makes sense only if the gender differences are obvious to the conversation participants [17. P. 15], i.e. gender isn't something people have — it's something they create [15. P. 707].
So, we have to admit, that the three basic types of contemporary textual analysis — narrative (biographical can be considered one of its varieties), discourse and conversational — differ by the focus of attention. The narrative analysis is an umbrella name of different approaches to the careful consideration of the specific type of texts, that is the so called narrative data as a combination of narrative and descriptive sentences, where we try to find basic structural elements paying attention to the linguistic nuances of the story as a way of personal and social identification construction here and now. No precise rules for empirical procedures within narrative analysis were written by any
of its adherents. To meet the requirements of validity and objectivity here we are to work out content analysis categorical scheme.
The same is the situation with discourse analysis that is an umbrella name of different interpretations of word 'discourse' itself and its relationships with the social reality (especially political and cultural practices). Discourse analysis adherents stick to different and usually conflicting theories, and sometimes the only thing that unites them is the absolute ignorance of what objects to study — they focus exclusively on the subject of identification, that is a specific discourse in the speech or real actions of some political, media, religious etc. persons, groups or communities. No precise rules for empirical procedures within discourse analysis were written by any of its adherents as well (we don't take into account some broad, general, vague and therefore frequently obscure recommendations given almost in any relevant article or a book). That's why "both conversation analysis inspired by ethnomethodology and discourse analysis are usually treated as self-sufficient approaches to studying the social world, rather than as mere methods that can be combined with others" [10].
So, again, to meet the requirements of validity and objectivity we are to work out content analysis categorical scheme within discourse studies as well. But it must be remembered that the content analysis, or textual coding, can be, on the one hand, pragmatic, semantic or psychological [21. P. 92; 18], on the other hand — quantitative and qualitative [26. P. 74—88]. Moreover we can't consider content analysis as a panacea in empirical work in the situation of terminological and methodological multiplicity of approaches for its 'heuristic inadequacy' [24. P. 199]: the whole coding scheme can be based on the previously established ideas and concepts helping the researcher just to illustrate or confirm his own point of view. Another source of mistakes and biases within any type of textual analysis — factors, that define the data 'truthfulness': factual emotions, communication processes and the human memory characteristics.
Conversational analysis had all the luck here — there are a lot of books and articles where mainly the same scheme with quite minor specifications is repeated, because the pretty stable terminology and ethnomethodological bases of conversational analysis have been set a long time ago. But the terminological confusions happen here too. For instance, some authors underlie differences between conversational analysis (represented by [14]) and critical discourse analysis (represented by [2]) in the light of post-structuralist discourse analysis that explores how fluctuating power relations between speakers are continuously reconstructed through competing discourses. This approach is an effective tool for explaining 'what is happening right now, on the ground, in this very conversation' and isn't concerned "with the modernist quest of seeking closure or resolutions in its analysis of what discourse means but rather with foregrounding the diverse viewpoints, contradictory voices and fragmented messages that research data almost always represents" [1].
So the post-structuralist discourse analysis "follows Foucault's thesis that the self is not fixed in a set of socialized, transferable roles, but is constantly positioned and repositioned through discourse. Individuals both negotiate and are shaped by their subject positions within a range of different and often conflicting discourses, which vary according to historical, cultural or social context. The motor for this, according to Fouca-
ult, is power, which is: never localized here or there, never in anybody's hands, never appropriated as commodity or a piece of wealth. Power is exercised through a net-like organization. And not only do individuals circulate between its threads; they are always in the position of simultaneously undergoing and exercising this power. They are not only its inert or consenting target; they are always also the elements of its articulation" [1]. In other words "the use of post-structuralist discourse analysis as an alternative methodology alongside those of conversational and critical discourse analysis that helps to challenge the inertia of 'linguistic orthodoxies' [3. P. 292] by opening up possibilities for new forms of writing that 'expose the self-interest and political economy of the sign 'critical'" [1].
Other authors directly insist that there are "two schools of contemporary discourse analysis — critical discourse analysis and conversation analysis" [4]. Moreover, starting from the observation that "critical trends in discourse analysis identify the intersection of language and social structure as the locus of critique", they qualify the treatment of context in some critical discourse analysis work as "largely backgrounding and narrative. Contextual information that invites critical scrutiny is often accepted as 'mere facts', framing the discourse samples analyzed in critical discourse analysis. On the other hand, context is reduced to a minimal set of observable and demonstrably consequential features of single conversations in conversational analysis and 'translocal' phenomena are hard to incorporate in conversational analysis analyses. Both treatments of context have severe defects... Using data from an ongoing project on narrative analysis of African asylum seekers' stories in Belgium, we discuss linguistic-communicative resources, 'text trajectories' (i.e. the shifting of text across contexts) and finally 'data histories' (i.e. the socio-historical situatedness of 'data')" [4]. So we see all the above mentioned pretty unclear terms in one small piece of article and the author uses them very voluntary.
The situation with identification of textual analysis type appears to be understandable and even predictable if we follow the distinctions stated before, in the beginning of the article — what is the focus of our attention: the object, the subject or the categorical scheme. The first two patterns leave us within the theoretical level of sociological work. That is, for instance, if we look for the dominant discourse that underlies biographical narratives of soviet people [30; 31] we can name our work both the narrative analysis (if we pay close attention to both the structural properties of texts and their subtle linguistic nuances focusing primarily on the tenuous complexity of actual analysis, concentrating on the object of our study, the narrative data itself) or discourse analysis (that supposes focus on the subject of study regardless the type of data under consideration for our crucial aim is the "powerful simplicity of reduction" [5. P. 94]). Let's be honest and confess that the first situation is possible only if the number of narratives/transcripts doesn't exceed 20—25 highly informative interviews. As soon as the number of such rushes to 30 and higher we have to admit the necessity and inevitability of reduction.
Nevertheless, regardless the data volume, we have to remember that "there is never a single message uniquely encoded in a text; there are several messages ("a net-
work of different messages") as decoded by different readers endowed with different 'intertextual frames' and 'intertextual encyclopedias', and different readings codes", "texts are hardly ever so 'closed' as to allow only one type of reading to the exclusion of all others" [9. P. 545—546]. We shouldn't also forget about empathetic understanding of the 'other' and sociological imagination that also distinguish researchers very much. Within the empirical sociological research you don't have right to humbly accept the inevitability of interpretations polyphony — the only appropriate way to work is to fulfill all the requirements to constructing content analysis categories and tools and do your job. How we and our colleagues prefer to call this process and its results — biographical, narrative, discourse, thematic, textual etc. analysis — doesn't really matter as long as you follow the scientific method rules (even the simplest frequency-counts of words have their value if properly used, backed up by other forms of evidence and being a relevant answer to the question asked).
REFERENCES
[1] Baxter J. Competing Discourses in the Classroom: a Post-Structuralist Discourse Analysis of Girls' and Boys' Speech in Public Contexts // Discourse & Society. — 2002. — Vol. 13(6). — Pp. 827—842.
[2] Billig M. Whose Terms? Whose Ordinariness? Rhetoric and Ideology in Conversation Analysis // Discourse & Society. — 1999. — Vol. 10. — Pp. 543—558.
[3] Billig M. Towards a Critique of the Critical // Discourse & Society. — 2000. — Vol. 11. — Pp. 291—292.
[4] Blommaert J. Context is/as Critique // Critique of Anthropology. — 2001. — Vol. 21(1). — Pp. 13—32.
[5] Chatman S. Story and Discourse: Narrative Structure in Fiction and Film. — Ithaka, NY: Cornell University Press, 1978.
[6] Cheek J. At the Margins? Discourse Analysis and Qualitative Research // Qualitative Health Research. — 2004. — Vol. 14. — N 8.
[7] Coates J. Women, Men and Language. — London: Longman, 1993.
[8] Crawford V.M., Valsiner J. Varieties of Discursive Experience in Psychology: Culture Understood Through the Language Used Selected for Their Divergent Viewpoints // Culture & Psychology. — 1999. — Vol. 5(3). — Pp. 259—269.
[9] Franzosi R. Narrative Analysis — or Why (and How) Sociologists Should Be Interested in Narrative // Annual Review of Sociology. — 1998. — Vol. 24. — Pp. 517—554.
[10] Hammersley M. Conversation Analysis and Discourse Analysis: Methods or Paradigms? // Discourse & Society. — 2003. — Vol. 14(6). — Pp. 751—781.
[11] Labov W. Language in the Inner City. — Philadelphia: University of Pennsylvania Press, 1972.
[12] Maynard D.W. On Co-implicating Recipients in the Delivery of Diagnostic News // P. Drew, J. Heritage (eds.) Talk at Work: Interactions in Institutional Settings. — Cambridge, UK: Cambridge University Press, 1992. — Pp. 331—358.
[13] Reed M. The Limits of Discourse Analysis in Organizational Analysis // Organization. — 2000. — Vol. 7(3).
[14] Schegloff E.A. Schegloffs Texts as Billig's Data: a Critical Reply // Discourse & Society. — 1999. — Vol. 10. — Pp. 558—570.
[15] Stokoe E.H., Weatherall A. Gender, Language, Conversational Analysis and Feminism // Discourse & Society. — 2002. — Vol. 13(6).
[16] Toolan M. Narrative: A Critical Linguistic Introduction. — London: Routledge, 1988.
[17] Tracy K. Analysing Context: Framing the Discussion // Research in Language and Social Interaction. — 1998. — Vol. 31(1). — Pp. 1—28.
[18] VincentR.C. A Narrative Analysis of US Press Coverage of Slobodan Milosevic and the Serbs in Kosovo // European Journal of Communication. — 2000. — Vol. 15(3). — Pp. 321—344.
[19] Barthes R. Miphologuii / Per. s fr. S.N. Zenina. — M., 2000.
[20] Dijk T.V. Jazik. Poznanie. Communikacija / Per. s angl. O.A. Guligui, sost. V.V. Petrova, pod red. V.I. Guerasimova. — M., 1989.
[21] Dmitrieva E.V. Focus-gruppy v marketingue i sociologuii. — M., 1998.
[22] Zolotova G.J. Communikativnaja grammatika russkogo jazika. — M., 1998.
[23] Isupova O.G. Conversacionniy analiz: predstavlenie metoda // Sociologuija: 4M. — 2002. — № 15.
[24] Kvadratura smysla: francuzskaya shkola analiza discursa / Per. s fr. u portug.; obsch. red. i vstup. sl. P. Serio; predisl. Ju.S. Stepanova. — M., 2002.
[25] Kupina N.A. Lingvistichiskiy analiz hudojestvennogo teksta. — M., 1980.
[26] Tarshis E.Ja. Perspectivy razvitija metoda content-analiza // Sociologuija: 4M. — 2002. — № 15.
[27] Urubkova L. Ponimanie i perevod // Vysshee obrazovanie v Rossii. — 2003. — № 5.
[28] Foucault M. Arkheologuija znanija. — M., 1996.
[29] Foucault M. Porjadok discursa. — M., 1996.
[30] Zvetaeva N.N. Biographicheskij discurs sovetskoj epokhi // Sociologuicheskij jurnal. —
1999. — № 1/2.
[31] Zvetaeva N.N. Biographicheskie narrativy sovetskoj epokhi // Sociologuicheskij jurnal. —
2000. — № 1/2.
[32] Shevchenko N. V. Osnovy lingvistiki teksta. — M., 2003.
ТЕКСТОВЫЙ АНАЛИЗ В СОЦИОЛОГИЧЕСКОМ ИССЛЕДОВАНИИ: ТЕРМИНОЛОГИЧЕСКИЕ ИЛИ МЕТОДОЛОГИЧЕСКИЕ
ТРУДНОСТИ?
И.В. Троцук
Кафедра социологии Российский университет дружбы народов ул. Миклухо-Маклая, 10/2, Москва, Россия, 117198
Проводятся демаркационные линии между различными вариантами текстового анализа. Все они обладают собственными теоретико-методологическими основаниями, которые тем не менее имеют массу точек соприкосновения, что наряду с терминологической путаницей и отсутствием четких инструкций по эмпирической работе с текстовыми данными затрудняет таковую в рамках социологического исследования.
Ключевые слова: текстовый анализ, нарративный анализ, дискурс-анализ, конверсационный анализ, лингвистический анализ, контент-анализ, кодирование.
