English for Specific Purposes 37 (2015) 74–86

Contents lists available at ScienceDirect

English for Specific Purposes

journal homepage: http: / /ees.e lsevier .com/esp/default .asp

Distinguishing textual features characterizing structuralvariation in research articles across three engineeringsub-discipline corpora

Budsaba Kanoksilapatham*

English Department, Faculty of Arts, Silpakorn University, Nakhon Pathom 73000, Thailand

a r t i c l e i n f o

Article history:Available online

Keywords:CorpusGenreDisciplinary variationEngineeringResearch articlesTextual organization

* Tel.: þ66 81 8584403.E-mail address: kanoksib@hotmail.com.

http://dx.doi.org/10.1016/j.esp.2014.06.0080889-4906/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Genre analysis has provided insights into the textual organization of different genres. Inthe research article (RA) genre, previous studies demonstrate that disciplinary variation isdiscernible. To raise the investigation to the level of sub-disciplines, this study addressestwo research questions: (1) What is the textual organization of individual RA sections incorpora from three engineering sub-disciplines? and (2) What are the significant statisticalvariations in textual organization that distinguish one engineering sub-discipline fromanother? Initially, three corpora were compiled, consisting of 180 full length high qualityRAs representing three sub-disciplines of engineering (civil, software, and biomedical).Then, the corpora were analyzed using genre analysis to identify the textual organizationprevalent in individual RA sections of each engineering sub-discipline. Subsequently, unitsof textual analysis called ‘move’ and ‘step’ were quantified and statistically analyzed tocapture significant statistical variations in each section. The analysis reveals the influenceof the sub-disciplines on the textual organization variations across the corpora, high-lighting the unique characteristics and perspectives of each sub-discipline. The findingscontribute to enhanced quality professional communication by creating and raisingawareness and sensitivity among prospective engineering students and practitioners whenthey are involved in the task of reading and/or writing RAs.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Currently, advanced degree students, university faculty members, and practitioners from diverse academic disciplines areencouraged to publish and disseminate their research discoveries in international journals. As a result, the expertise neededto be successfully involved in international professional communication is essential for academic and career growth. In thisregard, Swales’ genre analysis (1990, 2004), created initially to analyze the research article (RA) introduction section, has beenvery useful, providing a model that captures textual organization typically followed by RA authors. An increasing body ofresearch along this line has focused on other sections of the academic RA genre (e.g., Basturkmen, 2012; Kanoksilapatham,2005; Pho, 2008) and non-academic genres, including fund-raising letters (Connor & Gladkov, 2004), and recruitment ad-vertisements (Tisapramotekul, 2008).

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 75

A multitude of genre-based studies have been conducted on the genre of RAs and have provided valuable insight intothe rhetorical structure of the RA genre. However, certain criticisms have been raised against the application of genreanalysis. First, many genre-based studies have focused on individual RA sections (e.g., Basturkmen, 2012; Bruce, 2008;Kanoksilapatham, 2012; Lim, 2006; Peacock, 2011), resulting in fragmented knowledge of how this genre, in its entirety,is constructed. In addition, many studies have analyzed a small number of texts pertaining to individual sections of RAswhich were subjectively selected (e.g., Anthony, 1999; Basturkmen, 2009; Posteguillo, 1999), leading to limited general-ization of the findings. Thus, very little is known about textual organization across a large representative sample of textsfrom a genre. At the same time, corpus studies are currently revolutionizing the study of language use (Biber, Connor, &Upton, 2007). The analysis of large and representative text corpora can therefore provide a more rigorous and accuratedescription of textual organization that holds across all texts in the corpus (Kanoksilapatham, 2005; Stoller & Robinson,2013). Moreover, given the inherent nature of textual segmentation which is semantically driven and thus likely to besubjective, the integration of expert coders at the analysis stage can help validate move segmentation performed bydifferent individuals (Basturkmen, 2009, 2012; Stoller & Robinson, 2013). Finally, although studies have congruentlyrevealed the influence of disciplines on the internal discourse structure of each RA section, it remains to be investigatedwhether the influence of sub-disciplines associated with a single discipline can be observed in textual organization, and ifso, to what extent.

This study extends the application of genre analysis to examine the RA textual organization of three different engineeringsub-disciplines, by addressing the following two research questions: (1) What is the textual organization of individual RAsections in corpora from three engineering sub-disciplines? and (2) What are the significant statistical variations in textualorganization that distinguish one engineering sub-discipline from another? The answers to these two questions provide arepresentative description of how each RA section is structured. They also highlight the unique characteristics and per-spectives of individual sub-disciplines manifested in textual variations.

This article is structured as follows. First, an overview of Swales’ genre analysis and previous genre-based studies ondifferent RA sections is presented, as well as their shortcomings, which will be addressed by the current study in order tomake the analytical framework of genre analysis more solid and compelling. Next, the compilation and the analysis of threesizeable corpora representing prestigious and full length RAs from the three engineering sub-disciplines are described.Inter-coder analysis involving six experts was conducted to help validate the researcher’s textual demarcation. Then, thefindings generated from the genre analysis are presented, including the textual organization represented by a sequence of‘moves’ and ‘steps.’ Also, the frequencies of occurrence of individual moves and steps were compared and contrasted toobserve significant textual variations across the three corpora. Finally, certain limitations and implications of this study arestated.

2. Genre analysis

This section provides a brief introduction to Swales’ genre analysis – the analytical framework adopted by this study.Subsequently, previous genre-based studies are reviewed, particularly those on the four RA sections. Concurrently, drawbacksof the application of the framework are highlighted.

2.1. Swales’ genre analysis

The basic tenet of Swales’ genre or move analysis (1990, 2004) is that a text within a genre usually follows a typicalstructural pattern or organization, consisting of a series of moves sequenced in a particular order. A move refers to atext segment that performs a communicative function, contributing to the global function of a whole text. Moves canvary in length, but normally contain a proposition (Connor & Mauranen, 1999) and can be recognized by a set oflinguistic features. The flexibility of the model is made possible by incorporating the obligatory versus optional statusof the steps, based on the frequency of occurrence of each move. Finally, the model introduces the notion of the‘cyclicity’ of moves, referring to a situation in which a move can recur within a single introduction. To help contributeto the function of individual moves, each move may, in turn, contain multiple sub-units or a combination of sub-unitscalled ‘steps.’

The analysis of RA introductions from diverse academic disciplines led Swales to propose a three-move model in 1990,which was revised in 2004 to capture the variations of RA introductions reported. Swales’ (1990 and 2004) models for RAintroductions are generally similar, consisting of three principal moves:Move 1: Establishing a territory, Move 2: Establishing aniche, andMove 3: Presenting the present study. Move 1 introduces the research topic, Move 2 identifies the specific areas thatrequire further investigation, and Move 3 introduces the current research study highlighting certain prominent features ofthe research being presented.

In turn, each move can accomplish its function by a number, or combination, of steps, sub-units of a move. For instance, inSwales’ (2004) model, Move 2 can be realized by three possible steps including Step 1: Indicating a gap, Step 2: Adding to whatis known, and Step 3: Presenting positive justification. Similarly, Move 3 can be accomplished by as many as seven steps. Asopposed to Move 2, Moves 1 and 3 are quite frequent. Finally, as stipulated in the model, Moves 1 and 2 tend to be cyclical orrecursive, particularly in longer introductions.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8676

2.2. Previous genre-based studies on the four major RA sections

With reference to Swales’ introduction framework (1990, 2004), genre-based studies are numerous, sharing the commongoal of identifying textual organization. In this section, the application of genre analysis on the RA genre is scrutinized, and theareas where genre analysis falls short are identified.

Genre analysis, as a discourse framework, is inherently semantically driven because text segmentation is determined byhow an analyst interprets its communicative function. Thus, individuals, when analyzing the same segment of texts, canpossibly have different opinions with regard to its communicative function because different definitions are assigned to amove and step. For example, Kanoksilapatham (2005) considered Procedures and Procedural justification as two distinct movesin the results section, which might be considered as one single move by other researchers. Therefore, to help verify theoutcomes and validate the analysis, genre analysis should be complemented by an additional inter-coder procedure (e.g.,Basturkmen, 2009, 2012; Kanoksilapatham, 2005, 2007a; Stoller & Robinson, 2013). This procedure, including coder trainingand the actual coding of a portion of a corpus, is strongly encouraged, particularly when a researcher is not a member of thediscipline or discourse community and/or a researcher is not familiar with, or has limited knowledge of, the topics. Asatisfactory level of agreement reached by this inter-coder analysis can help confirm that the textual analysis is accurately andreliably conducted.

A number of studies focus on individual sections of RAs (e.g., Basturkmen, 2012; Bruce, 2008, 2009; Kanoksilapatham,2011a, 2011b, 2012, 2013; Lim, 2006; Ozturk, 2007; Peacock, 2002, 2011; Posteguillo, 1999; Williams, 1999) and a com-bined section of, for instance, results and discussion (Yang & Allison, 2003). Although the focus on individual sections allowsus to have a better knowledge of how particular RA sections are constructed, in order for advanced degree students, facultymembers, and practitioners in engineering to survive and thrive, the fragmented knowledge of each section does notcontribute much in this regard. Moreover, as required by genre analysis, in order to accurately understand RAs in their en-tirety, it is crucial that the researcher read the preceding and following sections to understand how the information is related.Therefore, studies that focus on full length RAs are beneficial for research community members (Kanoksilapatham, 2005,2007a; Lin & Evans, 2012; Nwogu, 1997; Stoller & Robinson, 2013).

A number of genre-based studies have analyzed relatively small datasets (e.g., Anthony, 1999 – 12 software engineeringintroductions; Basturkmen, 2009, 2012 – 10 discussions in language teaching and dentistry, respectively; Brett, 1994 – 20sociology introductions; Lim, 2006 – 20 business management methods). The application of genre analysis on a small datasetrenders the findings questionable and ungeneralizable. These findings thus remain to be substantiated by the analysis of asizable corpus. In this regard, the integration of the corpus notion with genre analysis is likely to provide us with moreconvincing and generalizable results. For instance, Kanoksilapatham’s (2005 and 2007a) studies on biochemistry andmicrobiology and Stoller and Robinson’ (2013) on chemistry, each with a corpus of 60 RAs, seem to provide assurance in thefindings generated.

With relevance to the size of a corpus, the representativeness of the corpus must be taken into consideration. For instance,Peacock (2011) analyzed 288 RA methods of eight disciplines taken from different journals nominated by two facultymembers of each academic department and published during the years 2002–2003. Bruce (2008) analyzed RA methodssections from the journals nominated by academics from one university and medical staff from one hospital in New Zealand.Posteguillo’s (1999) study examined 40 RA introductions in computer science selected from three different academic journalsrecommended by university teachers in Spain. Similar selection procedures have been adopted by many studies (e.g.,Anthony,1999; Basturkmen, 2009, 2012; Stoller & Robinson, 2013). Journal nomination by individuals can be quite subjective,varying depending on their preference or familiarity. As a result, the representativeness of the corpora analyzed is ques-tionable. In an attempt to reduce subjectivity and enhance representativeness, an objective set of criterion in compiling acorpus was pursued by Kanoksilapatham (2005, 2007a). In her studies, based on the journal impact factors, the top five highquality journals were identified, providing the source of RAs to be analyzed. The RAs selected can be seen to represent highquality RAs in the discipline in focus. Subsequently, the textual organization that the analysis of representative corpora yieldscan be seen to represent the preferred organization of high quality RAs.

In short, previous genre-based studies have provided insights into a better understanding of textual organization.However, to strengthen the analytical framework, more attention should be paid to the stage of corpus compilation to makesure that it is sizable, representative, and covers the major RA sections. At the stage of genre analysis, the subjective nature ofthe analysis can be compensated for by integrating inter-coder analysis. For a valid comparison of textual organization acrossdisciplines, in order to accurately observe the impact of disciplinary variation, the move/step definition needs to becontrolled. Even so, textual differences identified can be a result of chance or the idiosyncrasy of individual authors. In thisstudy, with the use of statistical analysis, it is expected that significant statistical differences or substantial variations sta-tistically established can potentially be claimed as distinguishing features across disciplines, highlighting the impacts of sub-disciplines on the RA rhetorical structure.

Even though genre analysis has contributed substantially to a better understanding of how the RA genre is constructed, asHyland (2002) remarks, specific needs and demands for language use should be catered to. At this point, research in genreanalysis has, to a certain extent, accomplished its goal, illuminating the influence of disciplines on textual organization.However, given Hyland’s comments on specificity, a more intriguing and pertinent question emerges: Do RAs belonging todifferent sub-disciplines associated with a single academic discipline share the same textual organization? For example, sub-disciplines of applied linguistics include language acquisition and second language writing (Ozturk, 2007) and those of

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 77

psychology include clinical and health psychology and behavioral, cognitive and neural, social, developmental and educa-tional psychology (Martín & Pérez, 2014). Engineering as a discipline is quite diverse, encapsulating a multitude of sub-disciplines including civil engineering, chemical engineering, electrical engineering, mechanical engineering, industrial en-gineering, and so forth (Ward, 2009). It is thus of interest to investigate whether the RAs of different sub-disciplines areorganizationally identical.

As pointed out by one of the reviewers of a previous draft of this article, these sub-disciplines might be considered aconstellation of disciplines, rather than sub-disciplines. This article, following Ozturk (2007) and Ward (2009), highlights thegeneral coverage of the field (e.g., engineering), and the relatively more specific scope of its branches (e.g., civil engineering,chemical engineering, electrical engineering, mechanical engineering and industrial engineering). Therefore, the term sub-discipline is adopted in this study.

Multiple sub-disciplines associated with engineering, for example, are situated in various contexts, and involve groups ofpeople from different professional domains with diverse interests. It is therefore worth examining how these different sub-disciplines construct knowledge through RAs (i.e., how knowledge is formulated, transformed, and acknowledged as new).Although it cannot be assumed that all RAs in engineering follow the introduction–methods–results–discussion or IMRDformat, a large number of previous genre-based studies focus on individual sections or a combined section of IMRD. In orderto provide prospective engineering students and scholars with a headstart when undertaking the task of RA writing/readingin their disciplinary discourse, this study focuses on the analysis of the four principal RA sections from three sub-disciplines ofengineering: civil (CE), software (SE), and biomedical (BE).

3. Methodology

This section first introduces the two research questions addressed in this study. Next, the background of the three en-gineering sub-disciplines is presented. Finally, the details of the procedures adopted by this study are discussed. These includecorpus compilation, corpus analysis by genre analysis, inter-coder reliability analysis, textual organization formulation, andidentifying distinguishing textual features by multiple chi-square tests.

3.1. Research questions

This study focuses on the four macro sections of RAs (introduction, methods, results, and discussion) in three engineeringsub-disciplines, addressing two major research questions: (1) What is the textual organization of individual RA sections inthree sub-disciplines associated with engineering? and (2) What are the significant statistical variations that distinguishindividual sections of one engineering sub-discipline from another?

The first research question is motivated by the findings of previous genre-based studies of different sections of RAs thathighlight the influence of disciplinary variation on textual organization. By extension and to provide further levels of spec-ificity, this study investigates whether the RAs of the three sub-disciplines are structurally different. The second researchquestion is triggered by the need to capture specific textual features that differentiate the sub-disciplines.

3.2. Description of three engineering sub-disciplines

The three engineering sub-disciplines selected CE, SE and BE are characteristically distinct with regard to their history,course of development, and nature. First, CE is an engineering sub-discipline with a long history that dates back more than300 years to 1711 (Ferguson & Chrimes, 2011). This sub-discipline deals with the design, construction, andmaintenance of thephysical and naturally-built environment, including works like roads, bridges, dams, and buildings. Second, as a consequenceof the rapid development of technology, engineering has experienced dramatic growth, resulting in additional engineeringsub-disciplines including computer engineering and SE. SE emerged in the 1940s, and is now an established profession,focusing on the systematic application, maintenance, and development of computer programs (Glass, Vessey, & Ramesh,2002). Finally, BE is one of the fastest growing fields of technology which emerged as a new discipline in 1952. This sub-discipline emphasizes the application of engineering principles and techniques of engineering to solve biological andmedical problems and improve healthcare (Nebeker, 2002). Due to its interdisciplinary nature, it is of interest to observe theextent to which the fusion of biology and engineering exerts influence on the textual organization of the RAs in this sub-discipline.

3.3. Corpus compilation

In order to obtain valid genre analysis results, corpus compilationmust be carefully executed, ascertaining that the corporameet the basic requirements, including being sizable and representative of the genre. The top five high quality journals ineach of the three engineering sub-disciplines were identified. To minimize subjectivity and enhance the validity of theanalysis in terms of the prestige of the journals, the more objective criterion of journal impact factors was used.

Journal impact factors are widely recognized as one of the key indices of quality for scientific journals. Although methodsused to calculate the impact factor have been criticized (Hansson, 1995; Seglen, 1997), Saha, Saint, and Christakis (2003)successfully tested the impact factor’s association with journal quality as rated by clinical practitioners and researchers,

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8678

empirically verifying the impact factor’s validity as a quality measure for journals. In this study, the use of impact factorsrepresents a step in improving the representativeness of the corpora in engineering.

The top five journals for each sub-discipline based upon impact factors for 2005 (the latest ones available at the onset ofthe study) were identified. From each journal, 12 experimental RAs were randomly selected to ensure that they were evenlydistributed throughout the publication year of 2006. The length, research topic, and authors of the articles were not controlledin this selection. In summary, each sub-discipline was represented by a corpus of 60 RAs. The three corpora of 180 RAs yieldapproximately 1 million running words. A complete list of the journals consulted and the titles of the articles constituting thecorpora are provided in the section of Supplementary Materials.

Given that the purpose of this study was to identify the textual organization of individual sections of introduction,methods, results, and discussion, scrutiny of the three corpora was necessary to obtain the actual number of individualsections for subsequent analysis. The introduction section is the most straightforward, being easily recognized by the explicitheading of Introduction. All of the RAs in the corpora have an independent introduction section, except one RA in BE, resultingin a total number of 179 introductions to be analyzed by move analysis. The other three sections of methods, results, anddiscussion are diversely labeled and not always independent. For instance, the methods section is diversely labeled (e.g.,Experiment, Experimental runs and measurements, Study area and instrumentation, Materials and methods), and sometimescombined with the results section (e.g., Experimental setup and results). Similarly, the results section is either under otherlabels (e.g., Experimental setup and results) or combined with the discussion, whereas the discussion section is either com-bined with the results or conclusions section, or under other labels (e.g., Summary and conclusions). To maintain the focus ofthe study, only the four independentmacro sections were analyzed. The total numbers of these sections analyzed in this studyare reported in Section 4.1.

3.4. Genre analysis

To address the first research question (i.e., What is the textual organization of individual RA sections in three sub-disciplines associated with engineering?), individual RA sections were analyzed by genre analysis to identify the movesand steps used. Even though these two central units of genre analysis are qualitative in nature, the researcher’s reliance on thesame definition of moves and steps corresponding to their communicative functions facilitated the identification task acrossthe three corpora.

Then, to formulate the rhetorical structure of each section for each sub-discipline, the preferred sequence of moves andsteps demonstrated in the three corpora was identified. As remarked in Swales’ model, certain moves seem to occur morefrequently than others. Even though Swales (1990, 2004) made the distinction between the two classifications of a move(obligatory and optional), no specific criterion was offered. In this study, to determine the potential status of each move, thefrequencies of occurrence of individual moves were recorded. Based on Kanoksilapatham’s criterion (2005) and with somemodifications, in this study, a movewas classified as obligatory, conventional, or optional if found 100%, more than 60%, and inless than 60% of the corpora, respectively.

3.5. Inter-coder reliability analysis

Genre analysis in this study includes textual analysis by the researcher and inter-coder analysis that involved six experts(two from each of the three engineering sub-disciplines). Inter-coder reliability analysis was conducted to demonstrate that aunit of text can be defined in such a manner that different individuals can demarcate the boundary of text units at a suffi-ciently high agreement rate. Two university faculty members in each of the three sub-disciplines, with a PhD from an English-speaking country were invited to serve as coders. Following Kanoksilapatham (2005), for each RA section, the inter-coderprocedure included a two-hour training session for the coders, coding practice of the section in one RA, and independentcoding of the section in four RAs. For more details on this procedure, refer to Kanoksilapatham (2005). The percentageagreement rate by the two coders for each RA section of each sub-discipline was calculated.

3.6. Formulating textual organization

First, to address the first research question, the textual organization of each section is presented, delineating all moves andsteps. The genre analysis of the three corpora reveals a preferred pattern of moves and steps, leading to the organization ofindividual sections. The status of each move and step, based on its frequency of occurrence (in %), was classified intoobligatory, conventional, or optional. Finally, the cyclical patterning (if observed, when a move recurs) is reported.

3.7. Identifying distinguishing textual features

Genre-based studies have revealed that the textual organization of one discipline is distinct from another (e.g.,Kanoksilapatham, 2007a; Peacock, 2011; Stoller & Robinson, 2013). As a result, textual differences or variations found can beconsidered distinguishing features between academic disciplines. To address the second research question (i.e., What aresubstantial variations that distinguish individual sections of one engineering sub-discipline from one another?), the fre-quencies of occurrence of all moves and steps employed were analyzed by multiple chi-square tests to statistically establish

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 79

significant variations across the three engineering sub-disciplines. With a systematic control of the corpus design (withregard to size, representativeness, and coverage) and the use of the same set of definitions to identify moves and steps, thesubstantial differences pertaining to individual sub-disciplines and captured by chi-square tests can be claimed to representdistinguishing textual features across the three engineering sub-disciplines.

4. Results and discussion

This section reports on the total number of individual sections actually analyzed by genre analysis, followed by the resultsof the inter-coder reliability analysis. Next, to address the first question, the moves and steps identified in the sections ofintroductions, methods, results, and discussion are described. Then, the textual organization of each section is presented. Toaddress the second research question, substantial variations determined by chi-squares (p < 0.01) are highlighted anddiscussed.

4.1. Total number of texts for each section analyzed

To address the first research question of this study, only individual sections of introduction, methods, results, and dis-cussionwere analyzed. In this regard, the examination of the three corpora revealed that 179 RAs had independent and clearlymarked introduction sections (i.e., 60 from CE and SE each, and 59 from BE), 101 methods (32, 16, and 53 from CE, SE, and BE,respectively), 105 results (18, 27, and 60 from CE, SE, and BE, respectively), and 92 discussion sections (16, 39, and 37 from CE,SE, and BE, respectively).

4.2. Inter-coder reliability analysis

In general, high overall inter-coder reliability of identifying moves between the researcher and the coders across the foursections from each of the three engineering sub-disciplines was attained, ranging from 82.46% to 93.32%, with the highestaverage rate being for the introduction section (91.06%), followed by the discussion (86.17%), methods (83.42%), and results(83.32%).

4.3. Structural organization of the introduction section

Moves and steps constituting the textual organization of the three sub-disciplines introductions are shown in Table 1.Compatible with Swales’ (1990 and 2004) models, these introductions display the same sequence of three moves includingMove 1: Establishing a territory to provide background information of the research topic,Move 2: Establishing a niche to justifythe presence of the research being reported, andMove 3: Presenting the present study to introduce the current research. Move1 can be realized by a maximum of three steps including Step 1: Claiming centrality highlighting the importance of the

Table 1Textual structure and organization of introductions in three engineering sub-disciplines (N ¼ 179).

Move/Step CE (N ¼ 60) SE (N ¼ 60) BE (N ¼ 59) p-Value

Introduction Percent Introduction Percent Introduction Percent

Move 1: Establishing a territoryc 60/60 100.00a 60/60 100.00a 59/59 100.00a N/AStep 1: Claiming centrality 29/60 48.33 45/60 75.00 44/59 74.58 .002*Step 2: Making topic generalization 58/60 96.67 57/60 95.00 58/59 98.31 .606Step 3: Reviewing previous studies 56/60 93.33 44/60 73.33 58/59 98.31 <.001*Move 2: Establishing a nichec 43/60 71.67b 49/60 81.67b 51/59 86.44b .121Step 1: Indicating gaps 37/43 86.05 45/49 91.84 39/51 76.47 <.001*Step 2: Adding to what is known 8/43 18.60 8/49 16.33 11/51 21.57 .002*Step 3: Presenting positive justification 5/43 11.63 5/49 10.20 21/51 41.18 <.001*Move 3: Presenting the present study 60/60 100.00a 60/60 100.00a 59/59 100.00a N/AStep 1: Announcing purposes 47/60 78.33 49/60 81.67 49/59 83.05 .796Step 2: Summarizing methods 40/60 66.67 55/60 91.67 47/59 79.67 .003*Step 3: Announcing principal outcomes 27/60 45.00 50/60 83.33 27/59 45.76 <.001*Step 4: Claiming research values 23/60 38.33 44/60 73.33 16/59 27.12 <.001*Step 5: Outlining article structure 17/60 28.33 30/60 50.00 6/59 10.17 <.001*Step 6: Offering procedural justification 12/60 20.00 3/60 5.00 3/59 5.08 .007*Step 7: Clarifying terms 0/60 None 6/60 10.00 1/59 1.69 .01*Step 8: Describing study sites 9/60 15.00 0/60 None 0/59 None <.001*Step 9: Suggesting further research 0/60 None 1/60 1.67 1/59 1.69 .600

* ¼ Significant variation (p < 0.01).a Obligatory.b Conventional.c Cyclical patterning.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8680

research topic, Step 2: Making topic generalization presenting established general knowledge related to the topic, and Step 3:Reviewing previous studies contextualizing the study within the existing literature. Move 2 consists of three steps includingStep 1: Indicating gaps critically evaluating previous research, Step 2: Adding to what is known emphasizing additional insightalong the line of previous research, and Step 3: Presenting positive justification giving reasons why the current study is needed.Finally, Move 3 usually concludes the introductions, with a maximum use of nine steps. Not all of the nine steps of Move 3were used in the three sub-disciplines. For instance, Step 8: Describing study sites was employed only in CE introductions,whereas Step 9: Suggesting further research was found in SE and BE, not CE introductions.

Based on the frequency of occurrence of each move and step, across three sub-disciplines, Moves 1 and 3 are obligatory,being found in every introduction, whereas Move 2 is conventional with the occurrence rates ranging from about 72% to 86%.Move 1 usually begins the introduction section, followed by Move 2 (if used); Move 3 usually concludes the section. Moves 1and 2 are likely to be cyclical especially when the research reported is complex, addressing a set of research gaps or objectives.

Two steps of Move 1 represent distinguishing textual features among the three sub-disciplines. First, Move 1, Step 1:Claiming centrality was used at about the same rate of 75% in SE and BE, but was infrequently used in CE (only 48%). The lessfrequent use of this move/stepmay be the result of thematurity of CE. That is, as one of the oldest engineering sub-disciplines,the announcement of the significance of the fieldmight be unnecessary. Second,Move 1, Step 3: Reviewing previous studieswasfrequent in CE and BE (about 93% and 98%, respectively), but only moderately frequent in SE (about 73%). The lower rate of useof this step might be because SE is evolving rapidly to satisfy industry demands of changing expectations and competitivemarkets. Moreover, software development is so diverse that literature pertaining to particular topics may not be extensive oreven relevant (Ferguson & Chrimes, 2011).

Similarly, Move 2 displays three substantial variations at the step level across the three corpora. Move 2, Step 1: Indicatinggaps was frequently used in CE (about 86%) and SE (about 92%), but relatively frequent in BE (about 76%). Move 2, Step 2:Adding to what is known, as opposed toMove 2, Step 1, was much less frequent. This step was found in BE (about 22%), but lessfrequent in CE and SE (about 19% and 16%, respectively). Finally,Move 2, Step 3: Presenting positive justification, was used in CEand SE only about 12% and 10%, respectively, but 41% in BE. Because BE seeks to address medical challenges in order toimprove the quality of human health and life, a research study in BE must be meticulously justified to avoid disastrous orunpleasant effects on humans. For instance, implants have been developed for human enhancement. In this regard,biomedical engineers need to justify why a study on the use of implants in clinical trials or a study on the use of certainmaterials for implants is necessary before these devices go on the market.

Move 3 displays substantial variations in the frequencies of the five steps. Four out of five steps (Step 2: Summarizingmethods, Step 3: Announcing principal outcomes, Step 4: Claiming research values, and Step 5: Outlining article structure) weresignificantly frequent in SE. For instance, Step 3: Announcing principal outcomeswas found at a high rate of about 83% in SE, butonly about 45% and 46% in CE and BE, respectively. Similarly, Step 4: Claiming research valueswas frequently used in about 73%of the SE corpus, but only about 38% and 27% in CE and BE, respectively. These significant statistical differences highlight thepertinent characteristics of this sub-discipline. Specifically, the frequent use of Step 5: Outlining article structure in SE waspreviously reported by Posteguillo (1999) in computer science RAs. In this study, the frequency of this step in SE (50%) wasrecorded as twice as frequently as that in CE (about 28%) and five times as frequently as that in BE (only about 10%). A numberof explanations are possible. According to Glass et al. (2002), SE is a relatively new fast moving field with some journals beingpublished 15 times per year. Thus, the article outline, along with the other steps, potentially helps RA readers rapidly locatewhat they would like to read for further details, expediting the dissemination of research discovery.

Another plausible explanation lies in the inherent unique characteristic of research methods in SE (Ferguson & Chrimes,2011; Glass et al., 2002; Kitchenham, 2002; Shaw, 2001). Specifically, Glass et al. (2002) stated that SE is a research field thatexhibits a plethora of research methods or paradigms under the taxonomies of, for instance, mathematical and mathematicalproof (for studies utilizing mathematical techniques), simulation (for studies utilizing simulation as a research method),concept implementation (for studies whose research method is to demonstrate proof of a concept), laboratory experiment (forstudies comparing the performance of a newly proposed system with other existing systems), and exploratory survey (forstudies conducting an exploratory field study). Since each research method entails a set of research procedures or researchparadigms that might not be widely understood, RA authors may feel compelled to prepare prospective readers for the field’slarge research method repertoire.

The corpus-based analysis of the introductions is illuminating for a number of reasons. As summarized in Table 1, nosignificant statistical differences were found at the move level. In other words, the introduction section seems to be con-structed similarly across the three sub-disciplines, suggesting that, generally, these sub-disciplines share the same goal inintroducing their research topics. However, the analysis demonstrates that the identity of the sub-disciplines, as reflectedthrough multiple significant statistical differences, is imprinted in the introduction section. The findings, corroboratingBasturkmen’s (2012) observation, reveal that variations in textual structures can be captured at the step level. Basturkmen(2012) claimed that the motivations that trigger those variations include differences pertaining to the subject matter, aswell as the goals and aims of individual sub-disciplines. In addition to the above-mentioned factors, this study supportsBazerman et al. (2005), who propose that the impact of the sub-disciplinary contexts, including the objects studied, thenature of the focus of the research (be they human or not), and the pace of sub-disciplinary evolution, play a role in framingand addressing scientific inquiries and creating new knowledge, in turn shaping specific textual organization across sub-disciplines. It is thus interesting to explore whether the texts belonging to the other three sections of RAs in the threesub-disciplines are as textually distinct as those in introductions.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 81

4.4. Structural organization of the methods section

Table 2 displays a set of three moves in the methods section. They are Move 4: Describing procedures, Move 5: Featuringother methodological issues, and Move 6: Reporting and consolidating findings. Move 4 is the only one that is obligatory, with100% of occurrence across the three sub-disciplines. In contrast, Moves 5 and 6 are optional, with frequencies ranging fromabout 38% to 51%. Move 4 usually begins the methods section, followed by Move 5; Move 6 (if used) concludes the section.Moves 4 and 5 are likely to be cyclical.

Move 4 can be realized by seven steps including Step 1: Announcing objectives, Step 2: Specifying protocolized procedures,Step 3: Detailing procedures, Step 4: Providing procedural background, Step 5: Justifying procedural decisions, Step 6: Describingresearch sites, and Step 7: Declaring ethical statements. Not all of the seven steps of Move 4 were found in all three sub-disciplines; various combinations of steps across the three sub-disciplines were common in this section.

Two steps were statistically established to be distinguishing textual features of methods in three engineering sub-disciplines (Step 2: Specifying protocolized procedures and Step 4: Providing procedural background). Step 2 was absent in SE,but present in CE and BE (about 31% and 49%, respectively), suggesting that a number of scientific procedures have becomeestablished and familiar in CE and BE (e.g., the two-tank filing system in CE and two-step collagenase perfusion technique in BE).As a consequence, brief reference of the procedures is sufficient for readers to follow the precise steps entailed in individualprocedures. In contrast, the absence of this step in SE can be explained by the fact that SE is the field of study with multipleresearch methods and their variants to specifically address the diversity of research questions (Shaw, 2002; Johnson, Ekstedt,& Jacobson, 2012). Therefore, details of the methodological approaches employed, reflecting different types of inquiry of thefield, are provided to facilitate RA readers’ comprehension while reading the methods section.

Step 4, the other distinguishing feature among the three sub-disciplines in Move 4, is absent in CE but present in SE and BEabout 25% and 51%, respectively. Possibly, in the field of SE, the diverse nature of the research problems requires such a broadarray of research methods or paradigms that readers need to acquire information about the methods in order to betterunderstand and evaluate the results that are presented. However, in BE, the much higher rate of this step can be explained bythe interdisciplinary nature of the sub-discipline. That is, it is possible that some of the RA readers in BE are from thediscipline of either biology or engineering, and they might not be familiar with certain procedures presented in RAs. Forexample, in BE, engineers create biomedical intervention. However, it is doctors, surgeons, and biologists who are ultimatelymost interested in the outcomes. Because they are not engineers, it is possible that they do not understand the nuances ofengineering methodology. Therefore, to accommodate the needs of potential RA readers and facilitate the RA reading task,additional background information about procedure descriptions is provided.

Two steps of Move 4 (Step 6: Describing research sites and Step 7: Declaring ethical statements), although not statisticallyestablished as distinguishing features, deserve our attention. Step 6 was present only in CE (about 44%), but absent in SE andBE. Under scrutiny, Move 4, Step 6: Describing research sites of CE methods is similar toMove 3, Step 8: Describing study sites inCE introductions. The use of this move/step in the two sections of CE, in line with Peacock’s (2011) study of environmentalscience, is to provide detailed geographical characteristics of the location where the research was conducted. As mentioned,CE deals with the design, construction and maintenance of the physical built environment. Because environments vary

Table 2Textual structure and organization of methods in three engineering sub-disciplines (N ¼ 101).

Move/Step CE (N ¼ 32) SE (N ¼ 16) BE (N ¼ 53) p-Value

Methods Percent Methods Percent Methods Percent

Move 4: Describing proceduresc 32/32 100.00a 16/16 100.00a 53/53 100.00a N/AStep 1: Announcing objectives 23/32 71.88 9/16 56.25 43/53 81.13 .127Step 2: Specifying protocolized procedures 10/32 31.25 0/16 None 26/53 49.06 .001*Step 3: Detailing procedures 32/32 100.00 16/16 100.00 53/53 100.00 N/AStep 4: Providing procedural background 0/32 None 4/16 25.00 27/53 50.94 <.001*Step 5: Justifying procedures 25/32 78.13 12/16 75.00 33/53 62.26 .266Step 6: Describing research sites 14/32 43.75 0/16 None 0/53 None N/AStep 7: Declaring ethical statements 0/32 None 0/16 None 14/53 26.42 N/AMove 5: Featuring other methodological issuesc 12/32 37.50b 7/16 43.75b 27/53 50.94b .477Step 1: Describing materials and participants 11/12 91.67 7/7 100.00 22/27 81.48 .368Step 2: Setting apparatus 6/12 50.00 0/7 None 11/27 40.74 .076Step 3: Identifying data sources 4/12 33.33 1/7 14.29 11/27 40.74 .421Move 6: Reporting and consolidating findings 9/32 28.13b 7/16 43.75b 26/53 49.06b .162Step 1: Stating findings 9/9 100.00 6/7 85.71 26/26 100.00 .077Step 2: Interpreting findings 3/9 33.33 1/7 14.29 2/26 7.69 .166Step 3: Comparing findings 4/9 44.44 2/7 28.57 2/26 7.69 .042Step 4: Explaining findings 3/9 33.33 1/7 14.29 1/26 3.85 .061

* ¼ Significant variation (p < 0.01).a Obligatory.b Optional.c Cyclical patterning.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8682

drastically from place to place, a detailed and specific geographical description of study sites is indispensable for this type ofengineering research.

Finally, Move 4, Step 7: Declaring ethical statements was found only in BE (about 26%) and absent in CE and SE. Althoughother disciplines and RAs require that declarations be explicitly made when human subjects are involved in the research, theabsence of this step in the methods section in CE and SE indicates that BE is the only sub-discipline that conducts research onhuman subjects. As identified in the BE corpus, an approval from the Institutional Review Board and patients’ signatures oninformed consent documents are required, an expected practicewhen human subjects are involved. The use of this move/stepin the RA methodology section of BE might be the outcome of historical evolution or the professionalization of researchers infields that conduct experiments on human subjects. According to Enfield and Truwit (2008), an Institutional Review Board(IRB) is a committee designated to approve, monitor, and review biomedical and behavioral research involving humans.Historically, IRBs developed in North America in the 20th century as a response to unethical scientific research having beenconducted on human subjects in the health sciences. An IRB review assures that appropriate methodological procedures aretaken to protect the rights and welfare of humans participating as subjects in a research study. The employment of this move/step in BE thus demonstrates the enforcement of this measure and the compliance of the BE discourse community with thisethical research practice.

Move 5: Featuring other methodological issues consists of three steps including Step 1: Describing materials and participants,Step 2: Setting apparatus, and Step 3: Identifying data sources. Finally, Move 6: Reporting and consolidating findings displays theuse of four steps related to research results including Step 1: Stating findings, Step 2: Interpreting findings, Step 3: Comparingfindings, and Step 4: Explaining findings. No significant statistical difference across sub-disciplines was established in these twomoves and their constituent steps. However, the presence of Move 6 in this section highlights the communicative function ofboth reporting and consolidating findings, suggesting that the boundary between the methods and results sections isbecoming less rigid in these three sub-disciplines.

This study has demonstrated that the principal function of the methods section is to provide an account of the researchprocedures, as evidenced by the 100% occurrence of Move 4: Describing procedures. Meanwhile, the two significant statisticaltextual differences of Move 4, (Step 2: Specifying protocolized procedures and Step 4: Providing procedural background) focusingon procedure description, are quite revealing. The frequency of occurrence of these two steps indicates that CE researchprocedures are reasonably established (with a relatively higher rate of Step 2 and no use of Step 4). In contrast, SE (vice versafor the rates of the two steps) features a myriad of research procedures, entailing a wide range of methodological approachesto address the diverse nature of studies taken in the field. BE, a hybrid sub-discipline, features the use of both steps at aboutthe same rates of 50% and 51%, respectively. This finding highlights the interdisciplinary nature of the field. In short, althoughthe methods section seems to be monolithic to begin with, under scrutiny, the construction of this section demonstratesevolutionary paths of individual sub-disciplinary research procedures influenced by multiple factors, including the nature ofthe field, the goals, and the research activities employed to create knowledge.

4.5. Structural organization of the results section

Three moves were identified in the results section (Table 3). They are Move 7: Summarizing procedures, Move 8: Reportingresults, andMove 9: Commenting on results. Move 7 consists of four steps including Step 1: Briefing procedures, Step 2: Justifying

Table 3Textual structure and organization of results in three engineering sub-disciplines (N ¼ 105).

Move/Step CE (N ¼ 18) SE (N ¼ 27) BE (N ¼ 60) p-Value

Results Percent Results Percent Results Percent

Move 7: Summarizing proceduresc 15/18 88.89b 24/27 88.89b 51/59 85.00b .866Step 1: Briefing procedures 15/15 100.00 23/24 95.83 51/51 100.00 .249Step 2: Justifying procedures 3/15 20.00 8/24 33.33 27/51 52.94 .045Step 3: Defining terms 3/15 20.00 1/24 4.17 4/51 7.84 .221Step 4: Referring to previous studies 10/15 66.67 6/24 25.00 14/51 27.45 .011Move 8: Reporting resultsc 18/18 100.00a 27/27 100.00a 60/60 100.00a N/AMove 9: Commenting resultsc 17/18 94.44b 24/27 88.89b 53/60 88.33b .753Step 1: Interpreting results 15/17 88.24 19/24 79.17 46/53 86.79 .632Step 2: Explaining results 10/17 58.82 17/24 70.83 35/53 66.04 .726Step 3: Comparing results 11/17 64.71 10/24 41.67 24/53 45.28 .295Step 4: Exemplifying results 3/17 17.65 3/24 12.50 3/53 5.66 .293Step 5: Cautioning limitations 2/17 11.76 4/24 16.67 8/53 15.09 .908Step 6: Summarizing results 2/17 11.76 5/24 20.83 7/53 13.21 .632Step 7: Directing future research 2/17 11.76 2/24 8.33 1/53 1.89 .215

* ¼ Significant variation (p < 0.01).a Obligatory.b Conventional.c Cyclical patterning.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 83

procedures, Step 3: Defining terms, and Step 4: Referring to previous studies. Move 8 has no identified step, andMove 9 has sevensteps including Step 1: Interpreting results, Step 2: Explaining results, Step 3: Comparing results, Step 4: Exemplifying results, Step5: Cautioning limitations, Step 6: Summarizing results, and Step 7: Directing future research.

Move 7: Summarizing procedures andMove 9: Commenting on results are conventional in the three sub-disciplines, rangingfrom 85% to 95%. Move 8: Reporting results forms the central and obligatory move of this section. Move 7, if found, begins thesection, followed byMove 8. Move 9 concludes the section. These threemoves are cyclical, reporting on individual results or aset of results. Even though the principal function of the results section is to present major findings generated by the studybeing reported (Move 8), a high rate of Move 7: Summarizing procedures across the three sub-disciplines (about 89%, 89% and85% in CE, SE, and BE, respectively) is remarkable. Move 7 provides a preview of the research procedures in order to prepareand facilitate RA readers’ understanding of, and appreciation for, the results reported in the subsequentmove. The presence ofMove 9: Commenting on results in the methods section at a high rate (about 94%, 89%, and 88% in CE, SE, and BE, respectively)demonstrates that, similar to biochemistry (Kanoksilapatham, 2005) and different from chemistry (Stoller & Robinson, 2013),the results section of the three sub-disciplines not only report results but also provides a channel for situating current findingsin context, a venue for arguments to be framed and discussed, and a stage for additional investigation to be conducted.

Multiple chi-square tests conducted showed no significant differences in the frequencies of the moves or steps of thissection. In other words, as shown by this study, the results sections of the three sub-disciplines are constructed similarly,sharing the same textual organization and reflecting a similar goal across the three sub-disciplines in reporting the results.

4.6. Structural organization of the discussion section

The analysis of the discussion section reveals a list of three moves (Table 4). They areMove 10: Reviewing the present study,Move 11: Consolidating results, and Move 12: Stating limitations and future research. Moves 10 and 12 have no identified steps.Move 11 can be realized by a set of seven steps, most of them overlapping with the steps found in Move 6: Reporting andconsolidating findings of the methods section and those of Move 9: Commenting results in the results section.

The three engineering sub-disciplines display a sequence of three moves. Move 10 usually begins the section, followed byMove 11;Move 12 concludes the section. This finding is in agreement with that of previous studies on diverse disciplines (e.g.,Basturkmen, 2012 in dentistry; Holmes, 1997 in social sciences; Kanoksilapatham, 2005 in biochemistry).

Move 10 is conventional across the three sub-disciplines, at the rate of about 69%, 74%, and 95% in CE, SE, and BE,respectively. Move 11, the central and obligatory move of the section, is invariably present in CE, and slightly less frequent inSE and BE (about 92% and 95%, respectively). Move 12 has a range of frequencies from 69% to 85% in the three sub-disciplines.

Given the fact that the discussion section of these three sub-disciplines usually consists of a number of sub-sectionsfocusing on individual results or a set of results, recursion of Move 10: Reviewing the present study and Move 11: Consoli-dating results is common. This finding is compatible with that of other studies of several disciplines, including Peng’s (1987) inchemical engineering, Kanoksilapatham’s (2005, 2007a, 2007b in biochemistry and microbiology, respectively), andBasturkmen’s (2009, 2012 in applied linguistics and dentistry, respectively). The use of Move 10 in CE, SE, and BE (about 69%,74%, and 95%, respectively) suggests that, in general, authors from the three sub-disciplines tend to facilitate potential readers’gaining access to a snapshot of the study by including and positioning Move 10 at the beginning of the section. Thus, insteadof reading an RA in its entirety, readers can directly proceed to the discussion section to obtain a glimpse of what the entire RAis about.

Similar to the other three sections, no statistical differences were identified at the move level in the discussion sectionacross the three sub-disciplines. Again, this finding demonstrates how the argumentative discussion across the three sub-disciplines is similarly presented, revealing the larger engineering perspective shared by these three sub-disciplines. That

Table 4Textual structure and organization of discussions in three engineering sub-disciplines (N ¼ 92).

Move/Step CE (N ¼ 16) SE (N ¼ 39) BE (N ¼ 37) p-Value

Results Percent Results Percent Results Percent

Move 10: Reviewing the present studyc 11/16 68.75b 29/39 74.36b 35/37 94.59b .027Move 11: Consolidating resultsc 16/16 100.00a 36/39 92.31b 35/37 94.59b .520Step 1: Reporting results 16/16 100.00 31/36 86.11 35/35 100.00 .214Step 2: Explaining results 11/16 68.75 12/36 33.33 28/35 80.00 <.001*Step 3: Summarizing results 5/16 31.25 2/36 5.56 16/35 45.71 .001*Step 4: Interpreting results 15/16 93.75 22/36 61.11 33/35 94.29 .001*Step 5: Comparing results 10/16 62.50 2/36 5.56 25/35 71.42 <.001*Step 6: Exemplifying results 5/16 31.25 11/36 30.56 6/35 17.14 .355Step 7: Claiming values of results 4/16 25.00 11/36 30.56 15/35 42.86 .402Move 12: Stating limitations and future research 11/16 68.75b 33/39 84.62b 29/37 73.38b .411

* ¼ Significant variation (p < .01).a Obligatory.b Conventional.c Cyclical patterning.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8684

is, to convince the reader of the merits of the study results in engineering, the same set of communicative functions isemployed. Initially, the section justifies the presence of the current study, be it by way of the research questions/hypothesesposed or the literature review. Then, the section highlights major findings of the study by, for instance, contextualizing thecurrent findings and explaining any new understanding or fresh insights about the research problems. Finally, the discussionsection demonstrates how the study has moved the readers’ understanding of the research problem forward by statinglimitations and suggesting future research.

As far as the steps of this section are concerned, the seven steps of Move 11 were intertwined and combined in diversepatterns to construct an argumentative discussion. The three sub-disciplines share a common rhetorical structure in the useof three steps of Move 11. That is, Step 1: Reporting findings is central for this move, providing the basis for other steps todevelop across the three sub-disciplines, with the frequencies of occurrence ranging from 86% in SE to 100% in CE and BE,respectively. The two other steps that are frequently found across the three corpora include Step 6: Exemplifying results andStep 7: Claiming values of results.

Four out of the seven steps of Move 11 (Step 2: Explaining results, Step 3: Summarizing results, Step 4: Interpreting results, andStep 5: Comparing results) were statistically established as distinguishing features for the three engineering sub-disciplines,contributing to the uniqueness of SE with a low use of these steps. For instance, the use of Step 5: Comparing results wasinfrequent in SE at only about 5%, whereas in CE and BE it was as high as about 62% and 71%, respectively, demonstrating thatevaluative elements are a common strategy used in CE and BE in consolidating their findings. A pertinent question arises: whydo SE discussions differ from those of CE and BE, as far as the four steps of Move 11 are concerned? At this point, it should benoted that the presence of these four steps in the discussion section performs a communicative function of evaluating andcontextualizing results. To provide an explanation for the low occurrence rate of evaluative elements, a scrutiny of howresearch was conducted, reported, and discussed in SE is essential. For example, two studies from the SE corpus set out toimprove current practice. To accomplish the objective, a new technique was proposed in both studies. In the first study, tosupport that new technique, a tool was implemented and the success of the implementation was announced. In the secondstudy, the new technique was applied to the example of a toy. Then, the contribution of the study was claimed. As describedearlier, SE is the field that has experienced dramatic rapid growth and development. To accommodate this nature of the field,as shown in the two examples, the research outcomes seem to be prioritized.

In short, the analysis results of the discussion section seem to be in line with those of the other three sections of intro-duction, methods, and results. That is, globally, these four individual sections across the three engineering sub-disciplinesshare the common goals of introducing a research topic (introduction section), describing research methodologies(methods section), presenting results (results section), and discussing results (discussion section). Moreover, this studyhighlights the remarkable role of steps as distinguishing textual features of the three sub-disciplines associated with engi-neering. That is, each individual section of each sub-discipline is constructed differently, employing different sets of steps tocreate knowledge or an argument. These differences, as captured by steps in different moves pertaining to various sections,can be accounted for based on the authors’ goals and the nature of the studies belonging to sub-disciplines.

5. Conclusions

Amultitude of studies have asserted the role of disciplinary variation in academic discourse. This study has raised the roleof such variation to a finer level of sub-disciplines by demonstrating that the application of genre analysis can successfullyidentify the textual organization of the fourmacro RA sections from three engineering sub-disciplines.With the integration ofthe notion of corpus studies andmultiple chi-square tests, this study is able to capture crucial textual variations, all of them atthe step level. In this regard, the introduction section is the most textually diverse across the sub-disciplines, followed by thediscussion section and the methods section. The results section, however, seems to display homogeneity with regard to theemployment of moves and steps across sub-disciplines.

The awareness of, and the sensitivity to, the textual organization of individual sub-disciplines can facilitate and enhancethe success of professional international communication, be it academic reading or writing. Pedagogically, teachers of ESP canmake use of the knowledge generated from this study to equip their advanced degree engineering students with the skills ofreading or writing RAs in their discipline. In a similar vein, engineering faculty members and practitioners are likely to bemore successful, bearing in mind the global textual organization revealed by this study, when reading or writing RAs forpublication. Additionally, the classification of moves into three categories based on frequencies allow individuals who embarkon an academic endeavor to be aware that, given a set of moves and steps available for individual sections, they also havefreedom to manipulate the employment of certain moves and steps within the variations of the nature of their researchstudies. The findings of this study culminate in the conclusion that each sub-discipline is unique in nature, having a discoursecommunity with its own conventions and perspectives, which are manifested in the selective deployment of certain movesand steps. However, as the discourse of RAs evolves. so do disciplinary practices. Therefore, students, academicians, andpractitioners should always be alert to possible textual variations.

The findings of this study contribute to strengthening Swales’models of textual organization which features two levels ofmoves and steps. As illustrated in this study, the same set of moves was identified across the three sub-disciplines, high-lighting the common larger perspectives shared by the three sub-disciplines when approaching the RA genre in general andRA individual sections in particular. However, the steps identified in the models, as stipulated by Swales (1990, 2004), are notonly the sub-units of moves contributing to the communicative function of the moves to which they belong. In fact, as

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–86 85

demonstrated by this study, certain steps play a crucial role as distinguishing textual features characterizing individual sub-disciplines, depicting the impact of sub-disciplines in shaping how individual RA sections are constructed. The modelsencapsulating both moves and steps presented in this study thus provide engineering students and practitioners with aflexible and viable channel to marshal their ideas, which can then be manipulated in compliance with their disciplinarydiscourse.

The methodological procedures adopted by this study strengthen the analytical procedures entailed by genre analysis in anumber of ways. First, the integration of specialist inter-coders in verifying the textual segmentation task is insightful. Thisprocedural step might not be deemed essential when analyzing, for example, advertisements or movie reviews that can beeasily understood by laymen. However, in certain disciplines, for example engineering, which involves a certain level ofexpertise, this procedure becomes crucial, complementing the semantic-driven nature of genre analysis and contributing tothe reliability of the textual segmentation task. Second, the comparison of textual structures across disciplines has becomeattractive to genre analysts. For valid comparison, it is vital that the same set of definitions of moves and steps be employed.The integration of statistical analysis to capture significant differences or distinguishing textual features potentially mini-mizes the possibility that the differences captured are due to chance or idiosyncrasy.

In this study, the integration of corpus analysis to genre analysis has moved the project of genre analysis further along.Based on the two crucial tenets of corpus analysis in selecting the journals, the three sizable and representative corpora weresystematically compiled and subsequently analyzed. The precaution at this initial stage of corpus creation in this study allowsthe findings generated by this study to be more solid, more valid, more convincing, and more generalizable. The findings caneventually lead to a more accurate body of knowledge regarding how individual RA sections are structured.

At this juncture, one caveat is in order: given the size of individual engineering sub-disciplines, the research topics of onesingle sub-discipline might be diverse and possibly characteristically different. This study did not exert control of the topicsand thus the findings remain to be substantiated.

Acknowledgments

This research project was financially supported by the Thailand Research Fund (TRF), Grant No. RSA5080005. I would liketo thank two anonymous reviewers and the editor for their constructive comments on the earlier versions of the paper.

Appendix A. Supplementary material

Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.esp.2014.06.008.

References

Anthony, L. (1999). Writing RA introductions in SE: How accurate is a standard model? The IEEE Transactions on Professional Communication, 42(1), 38-46.Basturkmen, H. (2009). Commenting on results in published research articles and master’s dissertations in language teaching. English for Academic Purposes,

8(4), 241-251.Basturkmen, H. (2012). A genre-based investigation of discussion sections of research articles in dentistry and disciplinary variation. English for Academic

Purposes, 11(2), 134-144.Bazerman, C., Little, J., Chavkin, T., Fouquette, D., Bethel, L., & Garufis, J. (2005). Reference guides to rhetoric and composition. Parlor Press and WAC

Clearinghouse.Biber, D., Connor, U., & Upton, T. (2007). Discourse on the move. Amsterdam: John Benjamins.Brett, P. (1994). A move analysis of the results section of sociology articles. English for Specific Purposes, 13(1), 47-59.Bruce, I. (2008). Cognitive genre structures in methods sections of research articles: A corpus study. Journal of English for Academic Purposes, 7(1), 38-54.Bruce, I. (2009). Results sections in sociology and organic chemistry articles: A move analysis. English for Specific Purposes, 28(2), 105-124.Connor, U., & Gladkov, K. (2004). Rhetorical appeals in fundraising direct mail letters. In U. Connor, & T. Upton (Eds.), Discourse in the professions: Perspectives

from corpus linguistics (pp. 257-286). Amsterdam: John Benjamins.Connor, U., & Mauranen, A. (1999). Linguistic analysis of grant proposals: European Union research grants. English for Specific Purposes, 18(1), 47-62.Enfield, K. B., & Truwit, J. D. (2008). The purpose, composition, and function of an Institutional Review Board: Balancing priorities. Respiratory Care, 53, 1330-

1336.Ferguson, H., & Chrimes, M. (2011). The civil engineers: The story of the institution of civil engineers and the people who made it. London: ICE Publishing.

Retrieved from http://www.ice.org.uk/News-Public-Affairs/ICE-News/ICE-launches-The-Civil-Engineers–The-Story-of-theGlass, R. L., Vessey, I., & Ramesh, V. (2002). Research in software engineering: An analysis of the literature. Information and Software Technology, 44(8), 491-

506.Hansson, S. (1995). Impact factor as a misleading tool in evaluation of medical journals. Lancet, 346(8979), 906.Holmes, R. (1997). Move analysis and the social sciences: An investigation of the structure of research article discussion sections in three disciplines. English

for Specific Purposes, 16(4), 321-337.Hyland, K. (2002). Teaching and researching writing. London: Longman.Johnson, P., Ekstedt, M., & Jacobson, I. (2012). Where’s the theory for software engineering? Software, 29(5), 96.Kanoksilapatham, B. (2005). Rhetorical structure of biochemistry research articles. English for Specific Purposes, 24(3), 269-292.Kanoksilapatham, B. (2007a). Writing research articles in English: Microbiology. Journal of the Faculty of Arts, 29, 174-208.Kanoksilapatham, B. (2007b). Writing scientific research articles in Thai and English: Similarities and differences. Silpakorn University International Journal,

7, 172-203.Kanoksilapatham, B. (2011a). Reconciling efforts to understand engineering intellect: Research articles from linguistic perspective. International Journal of

Business and Technopreneurship, 1(1), 137-148.Kanoksilapatham, B. (2011b). Language of civil engineering introductions: Textual structure and linguistic characterizations. Asian ESP Journal, 7(2), 55-84.Kanoksilapatham, B. (2012). Structure of research article introductions in three engineering subdisciplines. The IEEE Transactions on Professional Commu-

nication, 55(4), 294-309.

B. Kanoksilapatham / English for Specific Purposes 37 (2015) 74–8686

Kanoksilapatham, B. (2013). Generic characterisation of civil engineering research article abstracts. 3L: The Southeast Asian Journal of English LanguageStudies, 19(3), 1-10.

Kitchenham, B. A. (2002). Preliminary guidelines for empirical research in software engineering. The IEEE Transactions on Software Engineering, 28(8), 721-734.

Lim, J. M. H. (2006). Methods sections of management research articles. English for Specific Purposes, 25(3), 282-309.Lin, L., & Evans, S. (2012). Structural patterns in empirical research articles: A cross-disciplinary study. English for Specific Purposes, 31(3), 150-160.Martín, P., & Pérez, I. (2014). Convincing peers of the value of one’s research: A genre analysis of rhetorical promotion in academic texts. English for Specific

Purposes, 34(2), 1-13.Nebeker, F. (2002). Golden accomplishments in BE. Piscataway, NJ: IEEE History Center. Retrieved from http://en.wikipedia.org/wiki/Biomedical_engineeringNwogu, K. N. (1997). The medical research paper: Structure and functions. English for Specific Purposes, 16(2), 119-138.Ozturk, I. (2007). The textual organization of research article introductions in applied linguistics: Variability within a single discipline. English for Specific

Purpose, 26(1), 25-38.Peacock, M. (2002). Communicative moves in the discussion section of research articles. System, 30(4), 479-497.Peacock, M. (2011). The structure of the methods section in research articles across eight disciplines. The Asian ESP Journal, 7(2), 99-122.Peng, P. (1987). Organizational features in chemical engineering research articles. In T. Dudley-Evans (Ed.), Move analysis and ESP (pp. 79-116). Birmingham,

England: University of Birmingham. ELR Journal 1.Pho, P. D. (2008). Research article abstracts in applied linguistics and educational technology: A study of linguistic realizations of rhetorical structure and

authorial stance. Discourse Studies, 10(2), 231-250.Posteguillo, S. (1999). The schematic structure of computer science RAs. English for Specific Purposes, 18(2), 139-158.Saha, S., Saint, S., & Christakis, D. (2003). Impact factor: A valid measure of journal quality? Journal of Medical Library Association, 91(1), 42-46.Seglen, P. (1997). Why the impact factor of journals should not be used for evaluating research. British Medical Journal, 314(7079), 498-502.Shaw, M. (2001). The coming of age of software architecture research. In The proceedings of the 2001 international conference on software engineering.

Retrieved from http://www.scs.cmu.edu/afs/cs/project/vit/ftp/pdf/shaw-keynote-rev.pdfShaw, M. (2002). What makes good research in software engineering? International Journal of Software Tools for Technology Transfer, 4(1), 1-7. Retrieved from

http://marian.shaw-weil.com/displaypaper.php?PAPER_ID¼11&header¼http://spoke.compose.cs.cmu.edu/shaweb/p/Stoller, F., & Robinson, M. (2013). Chemistry journal articles: An interdisciplinary approach to move analysis with pedagogical aims. English for Specific

Purposes, 32(1), 45-57.Swales, J. (1990). Move analysis: English in academic and research settings. Cambridge, England: Cambridge University Press.Swales, J. (2004). Research genres: Explorations and applications. Cambridge, England: Cambridge University Press.Tisapramotekul, O. (2008). Generic variability in English recruitment advertisements of Thai and British newspapers. Language Across Cultures, 1, 171-204.Ward, J. (2009). A basic engineering English word list for less proficient foundation engineering undergraduates. English for Specific Purposes, 28(3), 170-182.Williams, I. (1999). Results section of medical research articles. English for Specific Purpose, 18(4), 347-366.Yang, R., & Allison, D. (2003). RAs in applied linguistics: Moving from results to conclusions. English for Specific Purposes, 22(4), 365-385.

Budsba Kanoksilapatham is a Professor in the Department of English, Faculty of Arts, Silpakorn University, Thailand. Her research interests include Englishphonetics, sociolinguistics, and discourse analysis.