English for Specific Purposes, Vol. 16, No. 4, pp.

321-337, 1997
0 1997 The American University. Published by Elsevier Science Ltd
Pergamon All rights reserved. Printed in Great Britain
08894906/97 $17.00+0.00
PII: SOSSS-4906(96)00038-5

Genre Analysis, and the Social Sciences:

An Investigation of the Structure of
Research Article Discussion Sections in
Three Disciplines
Richard Holmes

Abstract-The Discussion sections of 30 social science Research Articles, 10

each from the disciplines of history, political science and sociology, were
analyzed according to a modified version of the moves, or communicative
categories, presented in previous studies. It was found that, although there
were fundamental similarities to the natural sciences, social science Dis-
cussion sections also displayed some distinctive features. History texts were
particularly distinctive, and of the three disciplines bore the least resemblance
to those of the natural sciences. It is suggested that this distinctiveness is
sufficient to justify the idea of a social science subgenre of the Research
Article genre and that data of this kind are needed if academic reading and
writing materials are to be developed that are sensitive to the structural
features of academic texts and in particular to the ways in which such features
vary according to discipline. 0 1997 The American University. Published by
Elsevier Science Ltd

Introduction

Since the early 80s applied linguists and language teachers, especially
those concerned with the teaching of ESP and EAP, have shown a great
deal of interest in genre-centered approaches to the analysis of written and
spoken discourse. This interest in genre analysis has, to a large extent, been
motivated by pedagogical concerns, in particular by the need to provide
satisfactory models and descriptions of academic and scientific texts and to
enhance the ability of non-native speaker students to understand and, where
appropriate, to produce them. It has been suggested, for example by Bhatia
(1993), that earlier studies of academic and scientific discourse have failed
to produce relevant and comprehensive accounts of such texts.

Address correspondence to: Richard Holmes, 23, Jalan Pukat Satu 19/23A, Seksyen 19, 40000 Shah Alam.
Selangor, Darul Ehsan, Malaysia (e-mail: richard@ppp.itm.my).

321
322 R. Holmes

The essence of the concept of genre, as now used in applied linguistics,

ESP and rhetoric, is an emphasis on the primacy of communicative purpose
and the ways in which communicative needs shape or influence both surface
form and deeper rhetorical structures. For example, reprint requests, which
Swales (1990) has discussed in some detail, can be considered members of
a single genre since they share a clear communicative purpose, the obtaining
of a copy of an article or paper, that results in their being structured in a
very distinctive fashion. A genre then can be briefly defined as a class of texts
characterized by a specific communicative function that tends to produce
distinctive structural patterns. The concept of genre has been discussed at
some length by several writers, including recently Devitt (1993)) Freedman
(1993) and Berkenkotter & Huckin (1993) but the most helpful definition is
still perhaps that provided by Swales (1990) :58) :

A genre comprises a class of communicative events, the members of which

share some set of communicative purposes. These purposes are recognized
by the expert members of the parent discourse community, and thereby
constitute the rationale for the genre. This rationale shapes the schematic
structure of the discourse and influences and constrains choice of content
and style. Communicative purpose is both a privileged criterion and one that
operates to keep the scope of a genre as here conceived focused on com-
parable rhetorical action. In addition to purpose, exemplars of a genre exhibit
various patterns of similarity in terms of structure, style, content and intended
audience.

There is now a substantial literature in English on a variety of academic

genres including abstracts, presentations, lectures, theses, dissertations and
textbooks. It is, however, the Research Article (RA) that has received the
most attention with several studies focusing on its overall structure. This is
scarcely surprising since in many disciplines the RA is the main channel of
scientific or scholarly communication, although its dominance has lately
been challenged by some social researchers (see Canagarajah 1996). Where
the structure of specific sections of the RA has been analyzed, attention
has been mainly directed towards the Introduction or, less frequently, the
concluding Discussion section. Swales 1981 study, which was the first full-
scale analysis of RA Introductions, has been particularly influential. The RA
Introduction has also been studied by Cooper (1985), Crookes (1986), and
Taylor & Chen (1991). Introductions to social science RAs have been ana-
lyzed by Dudley-Evans & Henderson (1990) and Holmes (1995). The central
portions of the RA, the Methods and Results sections, seem to be generally
regarded as relatively straightforward and unproblematic, although Conduit
& Modesto (1990) and Thompson (1993) indicate that this may not be the
case.
It is noticeable that little of the research by applied linguists has dealt
with the variability of discourse structures between or within disciplines.
This is a question that has been of some concern to scholars working in the
American tradition of rhetoric studies. Peck MacDonald (1987)) for example,
 Genre Analysis and the Social Sciences 323

has indicated that problems are formulated very differently in particular

disciplines, while Dillon (1991) argues that academic discourse is char-
acterized by considerable diversity, competition and contention.
This concern with disciplinary variability is part of a growing awareness
by rhetoric scholars and sociologists of science over the last two decades of
the intensely social nature of scientific and academic writing. Increasingly,
academic discourse is analyzed in the context of disciplinary communities
that are formed by a variety of social pressures and constraints. This
approach has been advocated with particular force by Bazerrnan (1993):ix):

We can no longer view writing as limited textual practice, understood only as

the bounded rules of the page. Nor is writing to be understood only as the
product of an isolated mind... Writing is potentially responsive to and depen-
dent on everything that is on the social stage...

Although there have been some studies of academic writing in the humani-
ties and social sciences, for example by McCloskey (1986)) Peck MacDonald
(1987,1990,1992), Dillon (1991)) Hunston (1993) and Brett (1994), the bulk
of the research to date on particular sections of the RA or on its overall
structure has dealt with the natural sciences. This is perhaps unfortunate
since a large and increasing number of non-native speaker students are
studying social science subjects through the medium of English. There is
therefore a pedagogical rationale for extending the genre analysis of the RA
into the social sciences. A further justification for studying social science
RAs is that this will enable us to determine how far the patterns observed in
the natural sciences are generalizable to all written academic discourse.
The objective of the present study is therefore to examine the structure
of the Discussion sections, or equivalent, of articles from the disciplines
of history, political science and sociology. Although there are studies by
M&inlay (1984)) Hopkins (1985) and Peng (1987)) research on this section
is limited compared to the Introduction and there appear, moreover, to be
no published studies to date that have focused on Discussion sections of
social science RAs. Given the variability of this section and the importance
attached to it by students and supervisors (Dudley-Evans 1986, 1993), the
absence of research is especially striking. This study will examine the ways
in which the Discussion sections of social science RAs differ from those in
the natural sciences and the ways in which there is variation within the
social sciences. Political science and sociology were chosen since they are,
by common consent, central social science disciplines. History was included
because its ambiguous status, “traditionally poised between the social sci-
ences and the humanities” as Dillon (1991:15) puts it, might help to dis-
tinguish the features specific to social science disciplines.

Methods

Thirty articles were selected for analysis, 10 each from the disciplines of
history, political science and sociology. The choice of journals was motivated
324 R. Holmes

by the need to control as much as possible for such variables as writers’

nationality, levels of experience and expertise, period of publication and the
special features of subdisciplines. Articles were selected from the American
Historical Review (AHR), The American Journal of Political Science (AJPS),
and the Sociological Quarterly (SQ) . These three journals are of comparable
status, are published in the United States by professional disciplinary associ-
ations (national in the case of the AHR, regional in the case of the AJPS and
the SQ) and show no obvious bias towards any subdiscipline or overt ideol-
ogy. The corpus was restricted to a period of two years to control for rapid
changes within any of the disciplines. As well as RAs, all three journals
publish a variety of items such as review essays, rebuttals, comments and
research notes. It is plausible that such texts represent distinct genres with
distinct communicative purposes and they were therefore excluded from
the corpus. Consequently, the corpus in the present study is restricted to
articles the primary function of which is to present original research. The
RAs in the corpus were selected by using a table of random numbers. Full
bibliographical details of the articles are given in the Appendix A In the text
of this study each item is also identified by a letter and a number.
Analysis of the RAs was based on a modified version of the model outlined
by Hopkins and Dudley-Evans (1988:118) for natural science Discussion
sections. They identity 11 moves that can be observed in Discussion sections,
although they found that only the second of these, Statement of Results, is
obligatory. The moves are: (1) Background Information, (2) Statement of
Result, (3) (Un)expected Outcome, (4) Reference to Previous Research
(Comparison), (5) Explanation of Unsatisfactory Result, (6) Exemplification,
(7) Deduction, (8) Hypothesis, (9) Reference to Previous Research (Sup
port), (10) Recommendation, and (11) Justification. For the present study,
a number of modifications were made, mainly by conflating moves or by
somewhat extending their scope. It was also found necessary to add one
new move, Outlining Parallel OYSubsequent Developments. This move was
found only in the concluding paragraphs of history articles and consists of
the presentation in summary form of data additional to that given in the
main body of the article. For example, the conclusion to RA H8, “Struggles
for the Screen,” the main body of which ends in 1929, contains a paragraph
briefly outlining events in the 1930s. Thus, the following list of moves was
adopted for the analysis of Discussion sections:
1. Background Information.
2. Statement of Result, or statement about the significance of the present
research.
3. Wnjexpected Outcome, in which the writer comments on whether the
result is expected or not.
4. Reference to Previous Research, in which the writer compares his or her
results with those reported in the literature or compares his or her
research procedures, objectives or assumptions with those of previous
writers or refers to previous work to support his or her generalization or
refers to a deduction or hypothesis generated by previous research.
 Genre Analysis and the Social Sciences 325

5. Explanation of Unsatisfactory Result, in which the writer suggests reasons

for a surprising result or one different from the results in the literature
or gives an example to support his or her explanation. This includes a
single case of a satisfactory result being explained.
6. Generalization, in which the writer makes a claim about the gen-
eralizability of the particular results or limits claims to generalizability or
raises, discusses and/or dismisses questions that are indirectly related
to his or her research results.
7. Recommendation, in which the writer makes suggestions for future
research or regarding public policy or justifies the need for such sugges-
tions.
8. Outlining Parallel OYSubsequent Developments, in which the writer sum-
marizes data from a period subsequent to the one covered in the main
body of the article or data about a closely related topic.

For present purposes, a move is defined as a segment of text that is shaped

and constrained by a specific communicative function. A more rigorous
definition can be found in (Nwogu 1991:114):

By the term “move” is meant a text segment made up of a bundle of linguistic

features (lexical meanings, propositional meanings, illocutionary forces, etc.)
which give the segment a uniform orientation and signal the content of
discourse in it. Each “move” is taken to embody a number of “Constituent
Elements” or submoves which combine to constitute information in the move.

In most cases the unit of analysis was the sentence. There is a case for
adopting a unit of analysis below the level of the sentence such as the clause,
phrase or T-unit since a small number of instances were observed where
writers embedded one move inside another or included two moves within a
single sentence, techniques that seem to be more prevalent in the natural
sciences. It was felt, however, that such a procedure would be too cum-
bersome for the present study and would not be worthwhile since it would
involve very few sentences. Where a sentence appeared to contain two
moves it was assigned to the move that seemed to be more salient. In the
very few cases where it was impossible to decide which of the two moves
within a sentence was more salient, it was coded as containing two moves.
This was necessary for only three sentences in the entire corpus. It must be
noted that this procedure does involve a degree of subjectivity that is perhaps
unavoidable. The main consequence of using the sentence as the unit of
analysis is that there is a slight underestimation of the degree of subtlety
and complexity in a few Discussion sections. It might be noted that Crookes
(1986) also argues for this procedure. If, however, we were to analyze a
single text or segment of text rather than comparing several texts, it might
be more appropriate to adopt for analysis a unit below the level of the
sentence. Table 1 below presents an example of analysis of a Discussion
section. The text is from history article H5, “Moving out and settling in:
326 R. Holmes

TABLE 1
Outline structure of discussion section of “moving out and settling in”

Paragraph Sentence(s) Move

l-2 2 Statement of Result

3-4 6 Generalization
5 4 Reference to Previous Research (reference to a hypoth-
esis generated by previous research)
6-9 6 Generalization
l-3 6 Generalization (including limitation of claims to gen-
eralizability)
4-5 7 Recommendation

residential mobility, home owning, and the public enfiaming of citizenship,

1921-l 9W The text is as follows:

Our study of residential mobility in interwar Riverside and the postwar United
States yields several findings. First, the evidence on tenure in Riverside before
1950, and especially before 1946, carries implications for understanding both
historical and postwar residential mobility. Historical estimates of 50 percent
turnover within a decade probably mask a cycle of movement that was much
shorter, on the average about two years. Also, interwar residential mobility
paralleled the general pattern of nineteenth-century mobility historians have
consistently found, which leads us to conclude that there was no fundamental
change in the American propensity to move until after World War II. Second,
the two major explanations advanced by postwar social scientists to explain
contemporary residential mobility cannot by themselves account for the
sudden, dramatic, and nationwide decrease in mobility that has characterized
the United States since 1950. We believe that Franklin Roosevelt’s New Deal
housing programs and policies ultimately constrained mobility by making
home ownership easier and more attractive than it had been since the mid-
nineteenth century. Roosevelt’s vision of a modernized social compact
depended on a residentially stable citizenry. The size and power of the state
that Roosevelt’s administration created allowed the federal government to
intervene directly and forcefully in the home-financing market to achieve the
president’s goal. It was not, however, simply a matter of the state providing
access to home owning that restrained the American people’s propensity to
move but rather its providing the means, the long term mortgage, which was
motivated by the government’s desire to promote better citizenship.

We recognize that documenting the change in residential mobility before and

after World War II does not explain it. We have tried to suggest that variations
in endemic socioeconomic characteristics do not appear powerful enough to
explain the dramatic shii in the historic pattern that we have observed.
Instead, any satisfactory explanation of the change in residential mobility
must take into account the enormous role of the state. William Leuchtenburg
recently argued that historians beset by myriad varieties of social history and
calls for synthesis needed to resurrect the state as a major factor in American
history. We concur and further suggest that historians will need to reevaluate
the role of Franklin Roosevelt’s New Deal state in restructuring such a basic
feature of American life as residential mobility, armed with social-scientific
sensibilities that will allow them to analyze the political articulation of social
 Genre Analysis and the Social Sciences 327

values, the bureaucratic implementation of social policy and the experience

of ordinary people. (1420)

The structure of this text seems fairly straightforward although there are
some points of difficulty. The reference to “findings” in the first sentence is
somewhat confusing since it seems to refer both to data uncovered by the
authors’ research and to deductions and hypotheses arising from the data.
The fifth sentence of paragraph 1 also presented some difficulty. At first
glance the sentence reads like an indication of a gap or weakness of the
kind normally found in Introductions. It was eventually classified as a Move
4 since it is unambiguously a reference to a hypothesis produced by previous
research although, unusually, it criticizes a previous hypothesis instead of
using it to support the authors’ generalizations. The fourth sentence of the
second paragraph was classified as part of a Move 7 since, although it refers
to another writer, it concerns a recommendation by that writer rather than
his previous research.

Results

The RAs in the corpus were first analyzed for general patterns of organ-
ization. It was observed that all the texts were divided into sections. Those
in the history articles were not titled and were demarcated by double spacing
and partial capitalization of the first line of each section, presumably as a
result of editorial policy. The sections in the political science and sociology
RAs, except in most cases for the Introductions, were titled although the
number of sections and their titles varied quite considerably.
The organizational patterns of the political science and sociology RAs
were quite similar. Firstly, they all contained a recognizable Introduction.
In 18 out of 20 cases this was untitled and was then followed by a titled
section. With one exception, this was an extensive section that dealt with
theoretical background, previous research and general topical information
in varying proportions. Political science and sociology RAs also had, with
two exceptions, sections in the main text that dealt with methods. One of
the exceptions discussed questions of method in an appendix. Two political
science and four sociology ras had fairly long sections in which hypotheses
were laid out. All of the RAs had identifiable sections dealing with Results
and Discussions.
Most of the history articles were rather different. All contained an ident-
ifiable Introduction and a Discussion section, or conclusion as the authors
would probably prefer to call it, but only two contained a section devoted to
background and only two a section that dealt with methods. They all had a
main section that, for the sake of comparability across the disciplines, will
be referred to as the Results section although normally historians would
probably call it an argument or a narrative.
It seems safe to say that there is a standard pattern of organization for
RAs in political science and sociology and perhaps for other social science
328 R. Holmes

disciplines consisting of Introduction-Background-Methods-Results-Dis-

cussion, with a number including a Hypotheses section between the Back-
ground and the Methods section. The presence of such lengthy Background
sections can perhaps be considered a distinctive feature of RAs in the social
sciences as opposed to those in the natural sciences and might reflect the
absence of an agreed theoretical framework and what Bazerman (1988:35)
calls “a fixed codified literature.” For history RAs, however, the preferred
structure is Introduction-a main argument or narrative section that for
convenience will be referred to as Results-Discussion or conclusion.
The distinctiveness of history texts is perhaps related to the discipline’s
concern with providing accounts of discrete events rather than with the
discovery of generalizable patterns. Extensive Background sections are
uncommon in history articles since the delineation of a larger context is
generally seen as unnecessary. Political scientists and sociologists do need,
as noted above, to establish such a context while it seems that natural
science Discussion sections tend towards brevity because much of the
context can be assumed to be common disciplinary knowledge. The absence
of a Methods section in most of the history texts is in all probability a
consequence of the nature of historical data which, while they may be
selectively ignored or emphasized, are not actually constructed by the
researcher.
The move structure of the Discussion sections (or Conclusions) was
analyzed using the model outlined earlier. The structure of the sections is
indicated below in Table 2. Thus the structure l-2-8-1 would indicate
that the Discussion began with Background Information, proceeded to a
Statement of Result and the Outlining of Subsequent or Parallel Developments
and concluded with the provision of further information. Move-units are
demarcated by the transition from one move to another. The number of
units therefore indicates the degree of structural complexity.
There is considerable variation regarding the moves included in the Dis-
cussion sections. It seems that no move is completely obligatory. The most
common moves are Move 6, Generalization, and Move 2, Statement ofResults,
but even these are omitted from four and six articles respectively. Other
moves that occurred in at least half of the sections analyzed are Move 1,
Background Information (observed in 15 RAs); Move 4, Reference to Previous
Research (20 RAs); and Move 7, Recommendation (17 RAs). Move 3, Wn)ex-
petted Outcome, was observed in 11 texts, Move 5, Explanation of (Un)sat-
isfactory Result, in four and Move 8, Outlining Parallel or Subsequent
Developments, in five. Disciplinary variation was observed with regard to
Background Information (present in four history, seven political science and
four sociology texts), Reference to Previous Research (four, nine and seven
texts) and Recommendation (four, five and eight texts).
The corpus was also analyzed to determine the frequency with which
particular moves were likely to open a Discussion section. Move 2, Statement
of Results, stands out very clearly as the preferred opening for Discussion
sections with 15 RAs choosing this option. The next most common openings
 Genre Analysis and the Social Sciences 329

TABLE 2
Move structure of RA Discussion sections

No. of
RA Structure (Moves) Move units

History
Hl Schweniger 2-3 2
H2 Childers 2-6-1-4-2-7-2+1-2-6-S 12
H3 Wortman 2-8-7-8 4
H4 Fuchs & Moth 2-6-4-6-4 5
H5 Tobey et al. 2-6-467 5
H6 Steams & Haggerty a68 3
H7 Goheen 4-l-4-6 4
H8 Ross 6-l-2-8-7 5
H9 Schwartz l-2-86 4
HlO Waldron 8 1
Political Science
Pl Roberts 1-4-141-4-1-6-3-5-6-l-6-7 14
P2 Jacobson l-4-26-7 5
P3 Morehouse l-3-2-5-3-4-3-5-2-3-5-4-6 13
P4 Rapoport et al. 2-4-2-3-4-5 6
P5 Studlar et al. 2-616-1-2-l-61-2-6-2 12
P6 Welch and Leege 6-4-6-2-3-~6-4~6-4-6-4-1-6-l-6 16
P7 Sniderman et al. 2-6-2-4-7 5
P8 Quinn & Shapiro 4-2-6-4-6-l-6-4-fX-7-4-2-1-2-1-2 17
P9 Feiock 64-7-&7-6-3-l-6-7 10
PlO Jenkins-Smith et al. 2-6-2-6-2-6-2-62-4-6 11
Sociology
Sl Wilson 2-1-3-1-2-64-2-7-6 10
s2 Levy 6--4-2-l-62-&242 10
s3 Hessler et al. 4-7 2
s4 Britton 2-3-6-2-6-I-242-&7 11
s5 Pescosolido l-2-1-4-&l-3-2-3-2-5-4-6-7 14
S6 Graham & Hogan 2-7-4-6-2-d-2-7 8
s7 Smith 2%67-67-&7 7
S8 Grasmick et al. 3-2-3-l-2-3-&7 8
s9 Hwang & Murdock 2-3-6-4-7 5
SlO Semyenov & Lewin-Epstein 2-6-2-6 4

were the provision of Background Information, which appeared five times

and Generalization, which was chosen four times. No other move was
observed as the opening move in more than three texts.
Turning to the closing moves of the Discussion sections, it was observed
that every move except for Move 1 made at least one appearance as a closing
move. The most frequent choice was Move 7, Recommendation, which ended
13 articles, followed by Move 6, Generalization, which occurred seven times.
It is noticeable that sociologists showed a strong tendency to select Rec-
ommendation as the closing move.
The number of sentences devoted to each particular move in the Dis-
cussion section was also calculated and is indicated in Table 3. It can be
seen that, as far as the amount of text given over to particular moves is
concerned, Move 2, Statement of Result, Move 6, Generalization, and Move
8, Outlining Subsequent 01 Parallel Developments, are the most prominent.
However, the prominence of Move 8 in the table is somewhat misleading
330 R. Holmes

TABLE 3
RA Discussion sections: number of sentences per move

Number of sentences per move (%)

Move History Political science Sociology Total

1 15 (4.63) 48 (14.37) 6 (3.125) 69 (8.12)

2 44 (13.58) 52 (15.57) 71 (36.98) 167 (19.65)
3 1 (0.31)’ 13 (3.89) 15 (7.81) 29 (3.41)
4 34 (10.50) 67 (20.06) 13 (6.77) 114 (13.41)
5 29 (8.68) 1 (0.52) 30 (3.53)
6 3: (10.49) 116 (34.73) 38 (19.79) 188 (22.12)
7 23 (7.10) 9 (2.69) 48 (25.00) 80 (9.41)
8 173 (53.40) 0 0 173 (20.35)
Total 324 334 192 850

since it appears only in history RAs and then in only half of them, although
when it does appear it takes up a substantial amount of text.
Variation according to discipline is highly significant with regard to Move
2, Statement of Result, which was favoured by sociologists, Move 6, Gen-
eralization, which was especially prominent in political science texts, Move
7, Recommendation, which was much more in evidence in sociology texts
than in the other two disciplines and Move 8, Outlining Parallel or Subsequent
Developments, which occurred only in history articles.
Overall complexity was another area where there was a marked difference
between the disciplines. History Discussion sections were the least complex
with 4.5move-units and political science the most complex with 10.9units.
Sociology occupied an intermediate position with 7.9 units. This is in marked
contrast to the Introductions where history R4s were the most complex (see
Holmes 1995). The writers of History Discussion sections also tended to
employ a more restricted repertoire of moves. On average, 3.3 moves were
observed in history Discussion sections, 4.7 in political science and 4.3 in
sociology. As well as being less complex and employing fewer moves, history
Discussion sections also accounted for a smaller proportion of the RA than
those in the other two disciplines.
The observation, referred to earlier, by Dudley-Evans (1986) that moves
tend to occur in a predictable order was found to be applicable to the social
sciences. In the present corpus, the moves higher up the list usually made
their first appearance before those lower down. There were, however, some
exceptions. Move 2 appeared before Move 1 several times and Gen-
eralization was observed to normally occur before either Move 4, Reference
to Previous Research, or Move 5, Explanation.
A few of the sections in the corpus had a straightfonvard linear structure
in which discussion proceeded from one move to another without recursion.
Thus RA S9, “Ethnic enclosure or ethnic competition: ethnic identification
among Hispanics in Texas,” opens the Discussion section with a Statement of
Results. This is followed by a comment on the expectedness of the outcome,
a generalization that includes a limitation on claims to generalizability, a
reference to previous research on another ethnic group and rec-
 Genre Analysis and the Social Sciences 331

ommendations for future research. Many texts, however, did not proceed in
such a linear fashion and were characterized by the recurrence of one or
more moves. A typical pattern was the appearance of Statement of Result or
Background Information followed by Generalization or Reference to Previous
Research. Sometimes there was quite an elaborate sequence of such cycles.
RA PlO, “Explaining change in policy subsystems: Analysis of coalition
stability and defection over time,” for instance, comprises five cycles, defined
here as segments of text beginning with a Move 1 or Move 2. The first four
cycles consist of Statement of Result and Generalization and the fifth of
Statement of Result, Reference to Previous Research, and Generalization.
Again, there were noticeable disciplinary variations. Only one history RA
was cyclical in that it contained more than one segment of text headed by a
Move 1 or Move 2. On the other hand, all the political science FL& comprised
at least two cycles and the mean number of cycles per section was 3.5.
Sociology was somewhere in between. Seven sections had a cyclical struc-
ture and the mean number of cycles was 2.7. An interesting feature of the
political science Discussion sections was that opening cycles tended to have
fewer moves than later cycles.
To review the ground covered so far, it seems that, although the model
derived from the analysis of natural science texts is, in its broad outlines,
applicable to the Discussion sections of social science RAs, the latter are
less predictable and there were noticeable disciplinary variations with regard
to complexity, patterns of cyclical organization, the choice of closing moves
and the presence or absence of particular moves.

Discussion

It appears, then, that the rhetorical structure of political science and

sociology Discussion sections is quite similar to that found in the natural
sciences in the sense that the same moves, or communicative categories,
are discernible, albeit in varying proportions and with a lesser degree of
predictability. One might also note the similarity of overall RA structure in
that the conventional four sections are usually present in social science RAs
and that the structure of the Introductions is broadly similar. Natural science
and social science RAs can therefore be regarded with a good deal of
confidence as members of the same genre. They are not, however, identical.
Political science and sociology RAs resemble each other sufficiently and are
sufficiently distinct from natural science RAs, as a comparison with previous
research indicates, to justify the proposing of a separate subgenre char-
acterized, as far as the Discussion section is concerned, by greater variability,
limited complexity and a smaller number of cycles than natural science RAs.
We have noted, for example, that there is no completely obligatory move
in social science Discussion sections. This is in marked contrast to Peng’s
(1987) study of chemical engineering Discussion sections, which found
that four moves, Information, Statement of Result, Comparison with Previous
332 R. Holmes

Result, and Deduction, occurred in all of the texts studied, although Hopkins
and Dudley-Evans (1988) found that only Statement of Result was mandatory
in biology and irrigation and drainage papers. It is also particularly striking
that a comparison of the moves observed in the social science texts with
those observed in Peng’s study, which provides a complete analysis of all
the texts, shows that generally it is either political science or sociology that
most closely resembles chemical engineering as far as the presence of
particular moves is concerned. This is consistent with the suggestion that
political science and sociology RAs belong to a subgenre that is similar but
not identical to that of the natural science RA while History RAs are rather
more distinct but still related.
It is also noteworthy that the chemical engineering Discussion sections
contained more cycles, defined as segments of text beginning with Move 1
or Move 2, and were more complex than their social science counterparts.
Those in Peng’s corpus contained a mean of 68cycles, significantly more
than the means of 3.5 and 2.7 observed for political science and sociology
Discussion sections. Chemical engineering Discussion sections contained a
mean of 21.9 move-units compared to 10.9 for political science, 7.9 for soci-
ology and 4.5 for history (Peng’s data has been reworked to conform to the
categories of the present study).
History RAs show particular distinctiveness on a number of points. In
terms of overall structure they usually lack a Methods section. The Intro-
ductions are very long and very complex while Discussions normally do not
have a cyclical structure and tend to be brief. It seems inappropriate,
however, to regard history RAs as a separate genre, since there is clearly a
communicative purpose shared with the other disciplines and since they
adhere closely to the Swalesian model in the Introductions. It is possible
that history articles may be regarded as a distinct subgenre, as part of a
humanities subgenre, or as combining features of social science and humani-
ties subgenres. Resolving this issue would require comparison with the
structure of articles from fields such as philosophy and literary and cultural
studies.
To recapitulate, it would seem, if we make some fairly bold assumptions
about the representativeness of chemical engineering, that social science
Discussion sections are less complex, less predictable and more likely to
employ a restricted repertoire of moves than the natural or hard sciences.
The comparative complexity of social science Introductions has been noted
by Crookes (1986) and Holmes (1995). It seems therefore that the social
sciences, in contrast to the natural sciences, tend towards complexity and
elaboration at the beginning of the RA rather than at the end.
It is plausible that such disciplinary variation reflects the tendency for
research in the 20th century to become more expensive, systematic and
hierarchically organized. Bazerman (1988) has discussed the process by
which the production of knowledge in the natural sciences has become
progressively more rational, cumulative and collaborative. Similarly, McNa-
mee & Willis (1994)) following Price’s (Price 1986) concept of “big science,”
 Genre Analysis and the Social Sciences 333

outline a trend towards bureaucratization that is objectively measured by

reference to quantitative data, collaborative authorship and external financial
support. The evidence of the texts used in the present study is that, by
these three indicators, political science and sociology are still rather less
bureaucratized that the natural sciences, while history is substantially less
so. It is not unlikely that bureaucratization has contributed to the increasing
conventionalization of genres, a suggestion that is supported by the lesser
predictability of the social science texts, particularly those from the least
bureaucratized discipline of history.
We might also note the relative scarcity in history of references to previous
research, a scarcity that is plausibly rooted in the limited development of
cumulative research programs and the absence of a theoretical consensus
in that discipline. This proposal is supported by the comments of a leading
American historian on the disintegration of the former disciplinary con-
sensus and the ultimate failure of the new social history of the 1960s to
provide a new one (Fischer 1989). Such references are more likely to be
found in political science and sociology texts, although in the latter they are
often very brief, but they are not mandatory as they appear to be in natural
science texts. It is also striking that a large section of the sociology texts,
well over half the total number of sentences, is devoted to Statement ofResult
and Recommendation. We might tentatively suggest that this results from
the prevalence of external sponsorship in that discipline and a perceived
need to respond to sponsors’ concern with the effective application of
research findings.
The corpus analyzed here is quite limited, being composed of thirty texts.
Further research might be profitably conducted within a single discipline to
determine the degree of variability according to subdiscipline, ideology,
region of origin and level of prestige. Until this is done, generalization about
particular disciplines will necessarily remain somewhat speculative.
The findings of this study may have some implications for the teaching of
EAP. Much of the recent discussion of the teaching of academic reading
and writing skills to non-native speaker students has been concerned with
the perceived inadequacy of available tasks, materials and syllabi, a
deficiency that may in part be a result of a lack of awareness of the role of
genre in academic discourse. Such unease was expressed as long age as
1984 @wales 1984) and as recently as 1995 (Thomas 1995). The need for
sensitization to the variety and complexity of academic discourse and per-
haps also to its social context is all the greater since many students approach
academic texts with inappropriate schemata that result from a failure to
grasp the characteristic functions and structures of genres.
It is likely that the production of appropriate and relevant materials and
syllabi for EAP/ESP courses requires an awareness of the range of genres,
the ways in which genres span disciplines and, equally, the ways in which
they vary according to discipline and perhaps even to subdiscipline. The
present study indicates that social science RAs display patterns that are
significantly different from those observed in natural science RAs. This kind
334 R. Holmes

of data might be of value if applied linguists are to have a full understanding

of the nature of academic texts.

(Revised version received 1 October 1996)

REFERENCES

Bazerman, C. (1988). Shaping written knowledge: The genre and activity of

the experimental article in science. Madison: University of Wisconsin Press.
Bazerman, C. (1993). Foreword to N. Blyler & C. Thralls (Eds.) Professional
communication: the social perspective. Newbury Park: Sage.
Berkenkotter, C., & Huckin, T. (1993). Rethinking genre from a socio-
cognitive perspective. Written Communication, 10, 475-509.
Bhatia, V. (1993). Analyzing genre: language use in professional settings.
London: Longman.
Brett, P. (1994). A genre analysis of the results section of sociology articles.
English for Specific Purposes, 13, 47-59.
Canagarajah, A. (1996). From critical research to critical research reporting.
TESOL Quarterly, 30,321-330.
Conduit, A., & Modesto, D. (1990). An investigation of the generic structure
of the materials/methods section of scientific reports. Australian Review
ofApplied Linguistics, Series S, No. 6, 109-134.
Cooper, C. (1985). Aspects ofarticle introduction in IEEEpublications. Unpub-
lished M.Sc. dissertation, University of Aston, UK.
Crookes, G. (1986). Towards a validated analysis of scientific text structure.
Applied Linguistics, 7, 57-70.
Devitt, A. (1993). Generalizing about genre: New conceptions of an old
concept. College Composition and Communication, 44, 573-586.
Dillon, G. (1991). Contending rhetorics: writing in academic disciplines.
Bloomington: Indiana University Press.
Dudley-Evans, T. (1986). Genre analysis: an investigation of the introduction
and discussion sections of M.Sc. dissertations. In M. Coulthard (Ed.),
Talking about text (Discourse Analysis Monographs No. 13, English Lan-
guage Research). University of Birmingham.
Dudley-Evans, T. (1993). Variations in communication patterns between
discourse communities: The case of highway engineering and plant
biology. In G. Blue (Ed.) Language, learning and success: studying through
English. London: MacMillan.
Dudley-Evans, T., & Henderson, W. (1990). The organization of article
introductions: Evidence of change in economics writing. In T. Dudley-
Evans & W. Henderson (Eds) The language of economics: the analysis of
economic discourse. London: Modern English Publications/British Coun-
cil.
Fischer, D. (1989). Albion’s seed: four British folkways in America. New York:
Oxford University Press.
 Genre Analysis and the Social Sciences 335

Freedman, A. (1993). Situating genre: a rejoinder. Research in the Teaching

of English, 27, 272-281.
Holmes, R. (1995). Genre analysis and the social sciences: an investigation of
the introductions, background sections and discussion sections of research
articles in history, political science and sociology. Unpublished M.A. dis-
sertation, University of Surrey, UK
Hopkins, A. (1985). An investigation into the organizing and organizational
features of published conference papers. Unpublished M.A. dissertation,
University of Birmingham, UK.
Hopkins, A., & Dudley-Evans, T. (1988). A genre-based investigation of
the discussion sections in articles and dissertations. English for Specific
Purposes, 7, 113-122.
Hunston, S. (1993). Professional conflict: disagreement in academic
discourse. In M. Baker, G. Francis and B. Tognini (Eds), Text and tech-
nology: in honor ofJohn Sinclair. Philadelphia: John Benjamins.
McCloskey, D. (1986). The rhetoric of economics. Brighton: Wheatsheaf.
McKinlay, J. (1984). An analysis of discussion sections in medical journal
articles. Unpublished M.A. dissertation, University of Birmingham,
UK.
McNamee, S., & Willis, C. (1994). Stratification in science: a comparison of
publication patterns in four disciplines. Knowledge: Creation, Dihsion,
Utilization, 15, 396-416.
Nwogu, K. (1991). Structure of science popularization: a genre-analysis
approach to the schema of popularized medical texts. English for Specific
Purposes, 10, 111-123.
Peck MacDonald, S. (1987). Problem definition in academic writing. College
English, 49, 315-331.
Peck MacDonald, S. (1990). The literary argument and its discursive con-
ventions. In W. Nash (Ed.) The writing scholar: studies in academic
discourse. Newbury Park: Sage.
Peck MacDonald, S. (1992). A method for analyzing sentence-level dif-
ferences in disciplinary knowledge making. Written Communication, 9,
533-569.
Peng, J. (1987). Organizational features in chemical engineering research
articles. English Language Research Journal, 1, 79-116.
Price, D. (1986). Little science, big science and beyond. New York Columbia
University Press.
Swales, J. (1981). Aspects of article introductions. Birmingham, UK: Uni-
versity of Aston Language Studies Unit.
Swales, J. (1984). Research into structure of introductions to journal articles
and its application to the teaching of academic writing. In R. Williams, J.
Swales and J. Kirkman (Eds), Common ground: shared interests in ESP
and communication. Oxford, UK: British CounciVPergamon.
Swales, J. (1990). Genre Analysis: English in academic and research settings.
Cambridge, UK: Cambridge University Press.
Taylor, G., & Chen, T. (1991). Linguistic. cultural and subcultural issues in
336 R. Holmes

contrastive discourse analysis: Anglo-American and Chinese texts. Applied

3, 319-336.
Linguistics,
Thomas, S. (1995). A discourse-orientated approach to ESP. RELC Journal,
25,94-115.
Thompson, D. (1993). Arguing for experimental “facts” in science: a study
of research article results sections in biochemistry. Written Communi-
cation, 10, 106-128.

Appendix

History
AI1 texts were published in the American Historical Review.
Hl Schweninger, L. (1990). Prosperous blacks in the south, 1790-1880. 95,
31-56.
H2 Childers, T. (1990). The social language of politics in Germany: the soci-
ology of political discourse in the Weimar Republic, 95, 331-356.
H3 Wortman, R. (1990). Rule by sentiment: Alexander II’s journeys through
the Russian Empire. 95, 745-771.
H4 Fuchs, R., & Moth, L. (1990). Pregnant, single, and far from home: migrant
women in nineteenth-century Paris. 95,1007-1031.
H5 Tobey Wetherell, R., & Brigham, J. (1990). Moving out and settling in:
residential mobility, home owning, and the public enframing of citizenship,
1921-1950.95, 1395-1422.
H6 Stearns, P., & Haggerty, T. (1991). The role of fear: transitions in American
emotional standards for children, 1850-1950. 96, 63-94.
H7 Goheen, R. (1991). Peasant politics?: village community and the crown in
fifteenth-century England. 96, 42-61.
H8 Ross, S. (1991). Struggles for the screen: workers, radicals, and the political
uses of silent film. 96,333-367.
H9 Schwartz, S. (1991). The voyage of the vassals: royal power, noble obli-
gations, and merchant capital before the portuguese restoration of inde-
pendence, 1624-1640.96,735-762.
HlO Waldron, A. (1991). The warlord: twentieth-century Chinese under-
standings of violence, militarism, and imperialism. 96,1073-1100.

Political science

AI1 texts were published in the American Journal of Political Science

Pl Roberts, B. (1990). A dead senator tells no lies: seniority and the distribution
of federal benefits. 34, 3138.
P2 Jacobson, G. (1990). The effects of campaign spending in house elections:
new evidence for old arguments. 34,334-362.
P3 Morehouse, M. (1990). Money versus party effort: nominating for governor.
34, 706-724.
P4 Rapoport, R., Stone, W., & Abramowitz, A. (1990). Sex and the caucus
participant: the gender gap and presidential nominations. 34, 725-740.
P5 Studlar, D., McAllister, I., & Ascui, A. (1990). Privatization and the British
electorate: microeconomic policies, macroeconomic evaluations, and party
support. 34,1077-1101.
 Genre Analysis and the Social Sciences 337

P6 Welch, M., & Leege, D. (1991). Dual reference groups and political orien-
tations: an examination of evangelically oriented Catholics. 35,28-56.
P7 Sniderman, P., Piazza, T, Tetlock P., & Kendrick, A. (1991). The new
racism. 35,423-447.
P8 Quinn, D., & Shapiro, R. (1991). Economic growth strategies: the effects of
ideological partisanship on interest rates and business taxation in the
United States. 35, 656-685.
P9 Feiock, R. (1991). The effects of economic development policy on local
economic growth. 35,643-655.
PlO Jenkins-Smith, H., St Clair, G., &Woods, B. (1991). Explaining change in
policy subsystems: analysis of coalition stability and defection over time.
35,851-880.

Sociology

All texts were published in The Sociological Quarterly

Sl Wilson, J. (1990). Public work and social participation: the case of farm
women. 31,107-121.
s2 Levy, E. (1990). Social structure and generations in art: A case study of the
Hebrew theater. 31,203-224.
s3 Hessler, R., Pazaki, S., Madsen, R., & Blake, R. (1990). Predicting mortality
among independently living rural elderly: a 20-year longitudinal study. 31,
253-267.
s4 B&ton, D. (1990). Homophobia and homosociality: an analysis of boundary
maintenance. 31.423-439.
s5 Pescosolido, B. (1990). The social context of religious integration and
suicide: pursuing the network explanation. 31, 337-357.
S6 Graham, L., & Hogan, R. (1990). Social class and tactics: Neighborhood
opposition to group homes. 31,513-529.
s7 Smith, S. (1991). Sources of earnings inequality in the Black and White
female labor forces. 32, 117-138.
S8 Grasmick, H., Bursik, R., & Cochran, J. (1991). Render unto Caesar what
is Caesar’s: religiosity and taxpayers’ inclinations to cheat. 32, 251-266.
s9 Hwang, S., & Murdock, S. (1991). Ethnic enclosure or ethnic competition:
ethnic identification among Hispanics in Texas. 32, 469-476.
SlO Semyonov, M., & Lewin-Epstein N. (1991). Suburban labor markets, urban
labor markets and gender inequality in earnings. 32,611-620.

Richard Holmes holds a Ph.D. in History from Binghamton University and

an M.A. in Linguistics (TESOL) from the University of Surrey. He is a
lecturer in Linguistics and EAP at Mara Institute of Technology, Shah Alam,
Malaysia and has previously taught at the University of Maiduguri, Nigeria
and the International Islamic University Malaysia.