3 the dynamic relationship between text and context
3.3 Macrothematic analysis
The third part of the methodology which will be presented in this chapter analyses the texts at macrothematic level. Professionals write and talk in accordance with the genre patterns which they have learnt and internalized as a part of their training and socializing into a com-munity. These patterns organize the content, both at micro and macro levels, in ways which are relevant for the activities within the par-ticular domain. For an analysis of how professional discourses vary and change over time, I have found it relevant to develop a method-ology which grasps the content organization at macro level. As micro and macro levels are interdependent, I will include literature on both levels below.
3.3.1 Theoretical background
The content organization of spoken and written professional dis-course – documents, meetings, negotiations, etc. – is to a great extent conventionalized. The progression of the information flow and the structure of argumentation and narration follow patterns which are formed within the particular professional community and internal-ized by the members of this community. Professionals write and talk in accordance with the genre patterns which they have learnt and internalized as a part of their training and socializing into a commu-nity. They have to learn how they should address their readers and listeners in a way that is both recognized as appropriate for the par-ticular purpose and can be understood by the participants in the com-municative event.
At a micro level, the information flow of texts has been described as an interchange between themes or topics (old information) and rhemes or comments (the new information). Czech textlinguists within the ‘Prague school’ related this information flow to a ‘func-tional sentence perspective’. They claimed that each element of a sentence has a communicative function within the sentence as a whole. The various elements of the sentence contribute more or less new knowledge, which makes it possible to assign different information values to them (Daneš, 1970; also cf. Enkvist, 1974;
Källgren, 1979).
Among the early textlinguists we also find scholars who endeavoured to develop a methodology for the description of the over-all structure of a text – its micro and macro structures (van Dijk, 1977).
In Kintsch and van Dijk (1978) textlinguistic and cognitive psych-ology theories are combined in a model which aims to grasp how a reader manages to create a mental representation of a text as whole, drawing on our stored knowledge of text patterns at macro level. The concepts of schema, script and frame (cf. 3.1.1), originally developed within artificial intelligence, turned out to have an explanatory force in relation to how we store knowledge of text patterns and text types and how they are used in writing and reading (Schank and Abelson, 1977; Thorndyke, 1977). By studying patients with brain injuries neu-rologists found how knowledge of wholes was stored in separate parts of the brain from knowledge of details (Lurija, 1976), findings which gave studies on holistic text patterns and text comprehensibility a cognitive basis.
Textlinguists interested in the macro structures of texts have also been influenced by narratologists who have tried to grasp the textual deep structures, e.g. Vladimir Propp, Roland Barthes and Algirdas Greimas. Other influential studies show how genuine popular narratives are steered by common patterns. When Labov and Waletzky (1967) interviewed poorly educated people they found the following common story structure: orientation, compli-cation, evaluation, resolution and coda. Grimes and Glock (1970) also analysed popular narratives to reveal a similarity between these story patterns and literary prose. Another pattern found in narrative is the ‘Problem-solution pattern’. In his book On the Surface of Discourse, Hoey (1983) distinguishes different general text patterns: a Problem Solution pattern, a Matching pattern, and a General-Particular pattern.
The aim of these theories developed within textlinguistics, cog-nitive psychology and narratology were to grasp universal linguistic patterns. For an analysis of professional discourse, however, it is neces-sary to find methods to grasp the more specific text features as well as those common to all texts. A relevant approach to the study of genre development emanates from a group of researchers in Birmingham, UK, in the 1980s.
In an article in 1981, Swales developed his famous CARS-model (CARS Create A Research Space) for the description of the intro-ductory parts of scientific articles. Swales analysed 48 articles from different academic disciplines, and found that the rhetorical structure of their introductions could be summarized in rela-tion to four ‘moves’, which most often appeared in the texts in
the following order:
CARS-model (Swales, 1981) Move 1. Establishing the field
a. showing centrality of the topic
b. stating current knowledge of the topic c. ascribing key characteristics
Move 2. Summarizing previous research Move 3. Preparing for present research
a. indicating a gap b. question raising
c. extending a finding Move 4. Introducing present research
a. stating the purpose
b. describing present research
Other researchers have made a similar rhetorical analysis of other parts of the texts, e.g. discussion and result sections (Adams Smith, 1990; Peng, 1987; Dudley-Evans, 1986). Dudley-Evans (1989) thus found the following categories relevant for the analysis of discussion sec-tions of scientific articles: background information, statement of result, (un)expected outcome, reference to previous research (compari-son), explanation of a surprising or unsatisfactory result, deduction, hypothesis, reference to previous research (support), recommendation, justification. Huckin (1987), who made an attempt to apply Swales’
four moves to the discussion section, found that there they appeared in reversed order:
Introduction Discussion
4. Introducing present research 1. Statement of major results 3. Preparing for present research 2. Redescription of gap 2. Summarizing previous research 3. Selective literature review 1. Establishing the field 4. Implication for larger issues
The CARS-model has been used for a number of studies. It has been elaborated on since the 1980s by Swales himself (1990) and many oth-ers. In his extensive studies Bhatia (1993) has developed rhetorical analysis as a tool for the analysis of text of different kinds.
3.3.2 Method for macrothematic analysis
The methodology I present below is inspired by the moves analysis used by Swales and Dudley-Evans. In order to make it useful for the analysis of texts from different periods and from different genres (scientific and popular science), I have modified and expanded the Birmingham model. The most important differences are a distinc-tion of a supertheme called conclusion, and of different macrothemes related to consequences and measures directed towards society.
This analysis of the content structuring comprises two steps. The first step involves a categorization of supertheme. Four superthemes were distinguished: Introduction, Theme development, Discussion, and Conclusion. The second step aims at a finer subcategorization of the content. For the Uppsala corpus of scientific and popular science articles from different periods (cf. Chapters 4 and 6), we found it use-ful to distinguish the following set of macrothemes. Although each macrotheme most often appears within one of the four superthemes, the two theme categorizations were made independently. Below, how-ever, I have chosen to list the macrothemes in connection to the super-theme to which they are most often related.
Supertheme Macrotheme
Introduction Presentation of the area
Problem/knowledge gap/
research question
Relevance of the problem/area
Research review/
research situation
Current situation
Earlier situation
Purpose of the investigation/
the treatment/the experiment
Theme development Phenomenon, e.g. disease, technology,
economic situation
Method
Implementation Results
Material
Follow-up or extension
Discussion Explanation/interpretation/analysis of results
Validation, i.e. data presented as evidence for conclusions or hypotheses Deduction, i.e. conclusion or
consequence of presented facts
Recommendation of new investigation, treatment or method
Summary of results
Expected or unexpected outcome/
answer or no answer to questions Hypothesis
Conclusion Future situation
External, societal consequences Unsolved or future problems External measures, i.e. measures intended to have societal impact Problems related to external measures Justification of recommended
investigation/treatment/measure Hypothesis of solution
Suggestion for future measure/action The macrothematic analysis of the corpus of scientific and popular sci-ence texts from different disciplines and periods presented above have included a categorization of each macrosyntagm, or clause, of the texts in terms of both supertheme and macrotheme. The variation and change in the content structuring of texts can thus be described through this analysis.
A superthematic categorization of the text parts can also be useful in providing a picture of the overall content structuring of the texts. As an example of how the linear progression of texts can be studied, I will present the method I used to analyse 90 scientific and popular science articles. Based on the results of the superthematic categorization of
Table 3.1 Superthematic structure in 90 articles (First published as Table 8.4 in Gunnarsson, 1993: 171)
Text Linear structure Text Linear structure
Science Popular science Note: E – economic articles, M – medical articles, T – technical articles
each article, I could describe its linear thematic progression in a way that made comparisons over time interesting. The four superthemes – Introduction, Theme development (T), Discussion (D) and Conclusion (C) were grouped together in three clusters or cycles: introduction, theme cycle and discussion cycle. Theme cycle here refers to a linear sequence starting with theme development and optionally followed by a discussion section and/or a conclusion section: the combinations TDC, TD, TC and T are each counted as one theme cycle. Discussion cycle refers to a linear sequence starting with the supertheme discus-sion and optionally followed by a concludiscus-sion section, the combinations DC and D are each counted as one discussion cycle.
Table 3.1 gives an exemplary picture of how such an analysis could be undertaken and how at an abstract level it could point to differ-ences in the content structure of texts from different periods and in different disciplines.
A comparison of the structures of the economic articles – E at the top of the table – with those of the medical – M in the middle – and the technical – T at the bottom – shows that the economic articles are of a more theme repetitive kind, while the medical and technical articles are more straightforward in character and begin with an introduction followed by one or two theme cycles. If we look at the medical and technical articles, we will further find that this straight, simple struc-ture is more characteristic of the later articles, period 3, than of those from periods 1 and 2. For medicine, we can also note that this linear structure characterizes science – left part of the table – rather than popular science – right part.
3.4 Conclusions
As mentioned earlier, this multidimensional methodology has been applied to several large text corpora: (1) to the ‘Uppsala LSP corpus’, which consists of 360 scientific and popular science articles pub-lished in Swedish journals and periodicals on economics, medicine and technology during three centuries; (2) to the ‘Uppsala contrastive corpus’, which consist of texts in English, German and Swedish, rep-resenting 14 different text types, produced within banks, structural engineering firms, university departments of history and occupational medicine in Great Britain, Germany and Sweden (in total 8,858 mac-rosyntagms); and (3) to Fredrickson’s corpus, which includes all the documents from 24 court cases – 12 from an American court of appeal (Michigan) and 12 from a Swedish court of appeal (Svea hovrätt).
In these different studies, the texts were manually analysed in detail, coded and processed by computer according to statistical methods.
For the scientific and popular science articles, the various analyses were also integrated in a comprehensive textual description. As the macrothematic analysis was carried out concurrently with the cog-nitive and the pragmatic analyses, it was possible to relate the cogni-tive text structure and the pragmatic illocution pattern to the various superthemes. This meant that we attained a broad and multifaceted view of genre variation and change, also in relation to context.
The Uppsala research programme on the emergence and develop-ment of scientific writing has also focused on vocabulary and ter-minology. An extensive study of how the vocabulary used in texts from various areas of expertise has changed and varied over time was carried out on the basis of a quantitative analysis of the entire LSP corpus. The aim of this quantitative study has been to map diachron-ically the vocabulary in scientific and popular science texts in eco-nomics, medicine and technology from 1730 to 1985. All 360 articles included in the corpus have been computerized and computer proc-essed in order to obtain frequency listings and concordances for the various text groups.
In different chapters of this book, I discuss results of analyses using the multidimensional text linguistic methodology presented above.
Chapters 4 and 6, which both are included in the next section on
‘Scientific discourse’, discuss the results of the multidimensional text analysis. Chapter 4, which deals with the socio-historical construc-tion of medical discourse, will also refer to the results of the quantita-tive analysis of the vocabulary of scientific texts.
In Chapter 11 included in the section on ‘Discourse in large busi-ness organizations’, the results of the application of multidimensional text analysis to texts produced in British, German and Swedish banks and structural engineering firms are discussed.
Notes
1. The ‘Uppsala LSP corpus’ was built up within two large research projects, which between 1986 and 1992 were funded by one of the major Swedish research founda-tions, namely HSFR. This Uppsala programme on scientific and popular scientific articles was undertaken at FUMS (Unit for Advanced Studies in Modern Swedish) at Uppsala University. The first, textlinguistic part of the programme, which stud-ied articles from three periods of the twentieth century, was carrstud-ied out within the research project ‘LSP texts in the 20th century’ and with a project team consisting of myself as director, Björn Melander, Harry Näslund and Björn Skolander. The second, vocabulary part, which studied articles from three centuries, 1730–1985, was carried out within the research project ‘The emergence of languages for spe-cific purposes’. The project team consisted of myself as director and Björn Skolander.
2. The ‘Uppsala constrastive corpus’ was built up within a research project entitled
‘Texts in European writing communities’. This project, which was funded by Riksbankens jubileumsfond (a major Swedish research foundation) between 1994 and 1997, was undertaken at FUMS, Uppsala University, and involved an interdis-ciplinary team from three departments of this university: the Scandinavian, English and German departments. The project team consisted of myself as director, Bo Andersson, Ingegerd Bäcklund, Anna Levin, Ulf Norberg, Lena Norling, Eva Danielsson and Marie Sörlin.
3. Kirstin Fredrickson built up the Swedish part of her corpus during the 1991–92 academic year, when she was a guest doctoral student at FUMS, Uppsala University.
In her doctoral dissertation in linguistics (Fredrickson, 1995), she compares appeal court documents from the Swedish Court of Appeal ‘Svea Hovrätt’ with documents from Michigan Court of Appeal.
This section, which comprises three chapters, explores the emergence and development of scientific discourse within medicine, technology and economics. Chapter 4, concerns the socio-historical construction of medical discourse. Medical articles from three centuries: the eight-eenth, nineteenth and twentieth centuries, are analysed and discussed in relation to three different scientific stages: the pre-establishment stage, the establishing stage and the specialized stage. The multidi-mensional methodology which was presented in Chapter 3 is used to analyze changes in text patterns at cognitive, pragmatic and macro-structural levels. The analysis also deals with changes in linguistic expressions of evaluation over time. The theoretical basis for the study is the constructivist framework developed in Chapter 2.
In Chapter 5, my analysis deals with the non-verbal representation in 90 scientific articles within technology, medicine and economics from the same period, 1730–1985. This chapter also concerns the con-struction of scientific discourse, in this case with a focus on graphic representation, formulas and tables. The study compares article from the three fields in order to find out if and how they vary as to the role of the non-verbal representation. The changes found over time are also discussed in relation to an assumption about a possible connection between scientific thinking and communication.
Chapter 6 views the development of scientific writing within eco-nomics from the perspective of internationalization and globalization.
I analyse how textual patterns changed when an originally national journal changed language from Swedish to English lingua franca and became an international journal with a global readership. My analysis focuses on the design of the journal, the general outline of its articles and the gradual changes in journal and article patterns. The develop-ment is discussed from the perspectives of the national scientific and linguistic communities.