Iberica 13 Ibérica 20 (2010): 9-34 ISSN 1139-7241 Abstract Biber et al. (1999) contend linking adverbials perform important cohesive and connective functions by signalling connections between units of discourse; however, there has been little previous corpus-based research in this important area of ESP. This paper describes an analysis of linking adverbials, such as “however” and “therefore”, in a corpus of 320 published research articles (RAs) across eight disciplines, four science and four non-science. New lists of linking adverbials were developed and the parameters of frequency, function and disciplinary variation were examined using WordSmith Tools. They were found to be more frequent than previously thought, with numerous statistically significant disciplinary differences, for example between the sciences and non- sciences. Also, they often clustered together in complex sequences. A close examination of RAs in two of the sciences revealed some reasons for the much lower rate of occurrence there. Authors developed claims in a different way, describing methods and results in a more narrative or descriptive style rather than explicitly telling readers the connections between ideas, claims and facts. Conclusions are that linking adverbials are more important in RAs as signalling and cohesive devices, and for helping RA authors construct and strengthen claims, than previously thought by experts in this field. Also, different disciplines achieve this in significantly different ways, confirming the importance of discipline variation when researching their use. Keywords: English for Specific Purposes, linking adverbials, corpus analysis, interdisciplinary research writing, genre analysis. Resumen El uso de adverbios conectores en art�culos acad�micos de ocho disciplinas diferentes Biber et al. (1999) afirman que los adverbios conectores realizan importantes Linking adverbials in research articles across eight disciplines Matthew Peacock City University of Hong Kong (China) enmatt@cityu.edu.hk 9 01 IBERICA 20.qxp 14/9/10 18:01 Página 9 Ibérica 20 (2010): 9-34 MATTHEW PEACOCK funciones cohesivas y conectivas señalando conexiones entre unidades del discurso; sin embargo, no existe mucha investigación basada en análisis de corpus en esta importante área de Inglés para Fines Específicos. En este artículo se analiza el uso de adverbios conectores, por ejemplo however y therefore, en un corpus de 320 publicaciones académicas de ocho disciplinas diferentes, cuatro de ciencias y cuatro no de ciencias. Se elaboraron nuevas listas de adverbios conectores y se examinaron su frecuencia, función y variación según las disciplinas utilizando WordSmith Tools. Se descubrió que su frecuencia era mayor de lo que se creía, produciéndose importantes diferencias entre disciplinas desde el punto de vista estadístico; por ejemplo, entre las ciencias y las no ciencias. Además los adverbios conectores aparecían juntos a menudo formando secuencias complejas. Un análisis más minucioso de los artículos de investigación en dos de las disciplinas científicas reveló algunas razones que justifican su menor frecuencia en dichos campos. Los autores desarrollaban su argumentación de forma diferente, describiendo los métodos y los resultados en un estilo más narrativo o descriptivo en vez de alertar explícitamente a sus lectores de las conexiones entre ideas, afirmaciones y hechos. Se concluye en este estudio que la importancia de los adverbios conectores en los artículos de investigación reside en su función como instrumento señalizador y de cohesión, así como en ayudar a sus autores a construir y a fortalecer su argumentación; y que dicha importancia parece mayor de lo que se pensaba. Además, en diferentes disciplinas esto se consigue de formas claramente diferentes, confirmándose así la importancia de la variación disciplinaria a la hora de investigar el uso de estas expresiones. Palabras clave: Inglés para fines específicos, adverbios conectores, análisis de corpus, artículos de investigación en diferentes disciplinas, análisis de género del discurso. 1. Introduction This paper describes a corpus-based analysis of linking adverbials such as “however” and “thus” in research articles (RAs) across eight disciplines –Chemistry, Computer Science, Materials Science, Neuroscience, Economics, Language and Linguistics, Management, and Psychology. The purpose of the research was to study interdisciplinary differences in the frequency, form and function of the linking adverbials that authors use. There seems to be little research in the area since Biber et al. (1999), apart from Biber (2006). In this article Conrad’s (1999: 3) definition of linking adverbials is used: 10 01 IBERICA 20.qxp 14/9/10 18:01 Página 10 “those adverbials that serve to connect two stretches of discourse”. Biber et al. (1999: 765, 875) add to this definition, saying they “make explicit the relationship between two units of discourse”. They signal these relationships and are therefore “important devices for creating textual cohesion”. The following extract from a Neuroscience research article in the corpus shows an example: The criterion measure, B, was positively skewed and thus log transformed values were used for all analyses. Thus here signals a logical connection – the author used log transformed values because measure B was positively skewed. In other words, the second unit is a consequence of the first. The RA was chosen for this study because of its importance for spreading knowledge. Hyland (1996) says RAs are a vital medium for legitimating findings and disciplines, and Williams (1998) calls them the preferred genre for communication among discourse communities, noting that the language of RAs defines these communities. Pressure to publish makes RAs very important for researchers, and therefore it is not at all surprising that authors wish to stress the importance of their work and persuade readers of the authenticity of their arguments and claims. Moreno (1997) calls this their primary aim. However, Hunston (1994) suggests that while some may see RAs as objective and impersonal, their real aim is to persuade. Hyland (2000) agrees, saying the main purpose of the RA is to persuade, convince peers, and establish credibility. The corpus was 320 published research articles, forty from each discipline. The research aim and approach was not only to explore these cohesive and connective functions, but to investigate how these adverbials may assist in the performance of other important functions in RAs – presenting, developing, and supporting claims; together with persuading, convincing peers, and establishing credibility. The disciplines were classified for this research as science or non-science. The following methods were used to achieve this. First, visits were made to the academic departments representing the disciplines, and experts asked to discuss the classification. There was little controversy regarding Chemistry, Computer Science, Materials Science, and Neuroscience (sciences), or Management and Language and Linguistics (non-sciences). Regarding Economics, however, experts noted that while they consider it a non-science, it is sometimes called a LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 11 01 IBERICA 20.qxp 14/9/10 18:01 Página 11 social science, and found in university Faculties of Humanities/Social Sciences. There was a little more controversy with Psychology. Sources said that while this topic has initiated some discussion, neither they nor their colleagues classify it as a science, adding that among the 15 or so branches of psychology, just one branch is considered a science: Clinical Psychology. Second, it is noted that university Psychology departments are not normally found in Faculties of Science. Therefore the decision was made to classify Psychology as a non- science for the present research. 2. Definition and functions of linking adverbials Conrad’s (1999) and Biber et al’s (1999) definitions of linking adverbials appear in the previous section. However, the terminology, definition, and functions of linking adverbials have occasioned further discussion in the literature, with other authors sometimes using varied terminology. Carter and McCarthy (2006) call them “linking adjuncts” and distinguish them from the closely related conjunction. Conjunctions like “because”, “neither”, “nor”, and “but” also link two clauses or other units, but differ from linking adverbials in that they have a “purely syntactic role” (Biber et al., 1999: 85). Biber et al. (1999) also discuss coordinating conjunctions, which they also call coordinators, and subordinating conjunctions such as “despite” and “in spite of ”, which they also call subordinators, noting that they are closely related to but different from linking adverbials. The authors further note that “though”, “so” and “yet” can function either as linking adverbials or as conjunctions – “though” as a subordinator, “so” and “yet” as coordinators. Their study provides an example of “though” functioning as a subordinator, “She had never heard of him, though she did not say so”. An example of “so” as a conjunction is “I rose early so that I’d get there on time”. The following extract from the corpus for this study shows an example of “so” functioning as a linking adverbial: “There are lots of different levels and angles of interest. So the decisions are slow and complicated” (Management).1 Biber et al. (1999) indicate adverbials fall into three different classes: (a) circumstance adverbials (e.g. “nowadays”), which add circumstantial information about propositions in clauses, (b) stance adverbials (e.g. “definitely”), which express stance towards clauses,2 and (c) linking adverbials, which serve a connective function, making the relationship between two units of discourse clear, and are important cohesive devices. MATTHEW PEACOCK Ibérica 20 (2010): 9-3412 01 IBERICA 20.qxp 14/9/10 18:01 Página 12 The authors further divide linking adverbials into six “semantic categories”, which express the following different relationships: 1. Enumeration (e.g. “first”, “second”) and addition (e.g. “also”). The latter mark the next unit of discourse as being additional. 2. Summation, such as “to conclude”. 3. Apposition, which show the following text is an example (e.g. “for example”) or reformulation (e.g. “that is”). 4. Result/inference (e.g. “therefore”), which show the following unit is a result, or a logical or practical consequence. It also marks the conclusions the reader is expected to draw, or connects claims to supporting facts. 5. Contrast/concession (e.g. “however”), which indicate alternatives. They add that some highlight contrasting information, often leading to the main point the writer wants to make, and others express reservations “about the idea in the preceding clause”. 6. Transition, for example “by the way”, which mean something is only loosely connected. 3. Previous empirical studies Few empirical studies seem to have used corpora to research linking adverbials: three important studies are Biber et al. (1999), Biber (2006) and Charles (2007). Biber et al. (1999) examined four registers in the Longman Spoken and Written English (LSWE) corpus: news articles, academic prose (book extracts plus RAs, 2.6 million words each), fiction and conversation. Academic prose was 5.3 million words. There were seventy-five book extracts, mostly technical trade books, from thirteen different disciplines, and RAs from fifteen different disciplines. Linking adverbials were much more common in academic prose and conversation than in fiction and news. Biber et al.’s (1999) estimations show that in academic prose the result/inference category was the most common, since in this type of text presenting and supporting claims is very important, and, as these are developed, there is a greater use of linking adverbials. Regarding stylistic preferences, their work demonstrates that choices for the result adverbials “therefore”, “thus” and “hence” show more variability by author, and that LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 13 01 IBERICA 20.qxp 14/9/10 18:01 Página 13 these three adverbials seem to be interchangeable. It is also noted that multiple adverbials can appear in a clause (e.g. “going slowly now”), and that a single sentence may contain more than one linking adverbial (e.g. “in addition” plus “however”). Biber (2006) presents corpus results for five individual linking adverbials in textbooks across five disciplines (760,600 words), in “institutional writing” (catalogues and student handbooks, 151,500 words), and in “written course management” (course packs and course management, 159,600 words). He reports linking adverbials to be less common in “written course management”, and still less common in “institutional writing”. Charles (2007) presents corpus results for eight individual linking adverbials in eight first language (L1) Politics M.Phil. theses (190,000 words), and eight L1 Materials Science doctoral theses (300,000 words). The author’s bar chart does not provide exact numbers, as in the case of Biber et al. (1999), but their estimated figures are approximately as indicated in Table 1, which shows the results from all three empirical studies. These three studies, referred to above, and particularly Biber et al. (1999), are by far the most important to date in this area of ESP; although Conrad, in a later paper repeating many of Biber et al.’s (1999) (see Conrad 2000), calls for much more research into linking adverbials, particularly regarding register variation. Semantic Category Biber et al 1999: LSWE acade-mic prose Individual Linking Adverbials Biber et al 1999Biber 2006 Charles 2007 Whole Semantic Category LSWE academic prose Business Textbooks Engineering Textbooks Humanities TextbooksNatural Science Textbooks Social Science Textbooks Politics M. Phil. Theses Materials Science Ph.D. Theses MATTHEW PEACOCK Ibérica 20 (2010): 9-3414 01 IBERICA 20.qxp 14/9/10 18:01 Página 14 LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 15 Ib é ri c a 2 0 ( 2 0 1 0 ): … -… 1 IS S N 1 1 3 9 -7 2 4 1 B ib e r e t a l 1 9 9 9 : L S W E a c a d e - m ic p ro s e B ib e r e t a l 1 9 9 9 B ib e r 2 0 0 6 C h a rl e s 2 0 0 7 S e m a n ti c C a te g o ry W h o le S e m a n ti c C a te g o ry In d iv id u a l L in k in g A d v e rb ia ls L S W E a c a d e m ic p ro s e B u s in e s s T e x tb o o k s E n g in e e ri n g T e x tb o o k s H u m a n it ie s T e x tb o o k s N a tu ra l S c ie n c e T e x tb o o k s S o c ia l S c ie n c e T e x tb o o k s P o li ti c s M . P h il . T h e s e s M a te ri a ls S c ie n c e P h .D . T h e s e s h o w e ve r 1 1 0 0 1 4 0 0 1 1 0 0 8 0 0 1 0 0 0 1 0 0 0 1 7 0 0 1 9 0 0 n e ve rt h e le ss 1 0 0 -- -- -- -- -- -- -- C o n tr a st / co n ce ss io n 1 2 0 0 ye t 1 0 0 -- -- -- -- -- 3 0 0 5 0 th u s/ th e re fo re 1 3 0 0 1 3 0 0 1 8 0 0 9 0 0 1 3 0 0 1 1 0 0 1 7 0 0 1 7 0 0 R e su lt/ in fe re n ce 3 0 0 0 h e n ce 1 0 0 -- -- -- -- -- 2 0 0 5 0 0 fu rt h e rm o re 1 0 0 -- -- -- -- -- 2 5 0 1 5 0 A d d iti o n 1 0 0 0 in a d d iti o n -- -- -- -- -- -- 1 0 0 3 0 0 fo r e xa m p le 6 0 0 1 4 0 0 1 4 0 0 7 0 0 1 0 0 0 1 0 0 0 5 0 0 3 0 0 e .g . 2 0 0 -- -- -- -- -- -- -- th a t is -- 4 0 0 9 0 0 2 0 0 3 0 0 3 0 0 -- -- A p p o si tio n 1 8 0 0 i.e . 2 0 0 -- -- -- -- -- -- -- T a b le 1 . P re vi o iu s e m p ir ic a l f in d in g s: F re q u e n cy p e r m ill io n w o rd s. 01 IBERICA 20.qxp 14/9/10 18:01 Página 15 4. Rationale for research As noted in the previous section, Biber et al. (1999) contend that presenting, supporting and developing claims, and textual cohesion, are very important in academic prose. Shaw (2000) also says it is important in RAs to develop arguments, and support claims. Shaw (2000) and Hunston (2001) both also point out that research writers organize arguments into long chains (also see Hoey, 2001), but neither author discusses how they do this. It is suggested that these authors are correct about the importance of supporting and developing claims in RAs, and that linking adverbials might play a role in helping authors to achieve this. Yet data on linking adverbials, across disciplines and particularly in the sciences, seems to be lacking. Liu (2008) calls for “more detailed” corpus-based research on linking adverbials, noting that this method is the only way to achieve accurate and reliable results. It is proposed that the area has not received the attention it warrants and that further research is needed, to assess disciplinary variation across a number of science and non-science disciplines. This research focuses on four of Biber et al.’s (1999) semantic categories for linking adverbials: “contrast/concession”, “result/inference”, “apposition”, and “addition”. The first three were chosen because the functions they perform appear to be an important part of presenting, supporting and developing claims. “Addition” was selected because preliminary examination of the corpus indicated that linking adverbials such as “also”, which Biber et al. say function to show the next unit is additional, in fact often aid writers to introduce claims. In order to make this research more manageable, other categories analysed by Biber et al. (1999) were left aside, because they may be less important in developing claims and are regarded as beyond the scope of the present study. Bhatia (2000) says a strong justification for genre research in ESP is that it informs the teaching of research writing, especially for writers who wish to join academic discourse communities. The area is increasingly important due to the fast-growing numbers of research writers around the world, and is worth further investigation. The results might tell us much more about ESP and the nature of RAs, and help teachers of research writing inform learners of appropriate patterns. Although there is a considerable amount of information about the RA available, much still remains to be discovered about this subject; and a number of authors have called for more research. Swales (2004) does not MATTHEW PEACOCK Ibérica 20 (2010): 9-3416 01 IBERICA 20.qxp 14/9/10 18:01 Página 16 contradict his previous claim that our picture of the RA “is far from complete” (Swales, 1990). Berkenkotter and Huckin (1995) note that understanding genres is vital for participation in relevant discourse communities. Hyland (1996 & 1997) says a better understanding of RAs will improve understanding of rhetoric and how scientists work. This will also help the teaching of ESP. Writing theses and RAs is difficult (Paltridge, 1993), and is a prerequisite for entry to the research community. Hopkins and Dudley-Evans (1988) suggest that ESP materials must be informed by genre research and that understanding disciplinary differences is essential for preparing ESP courses. The next section will explain the aims of this research and also describe the corpus, how the list of linking adverbials was built up, and how the corpus was searched. 5. Methodology The aims of this study were to advance and extend previous research on the form, frequency, function and distribution of linking adverbials in RAs across eight disciplines, four science (Chemistry, Computer Science, Materials Science, Neuroscience) and four non-science (Economics, Language and Linguistics, Management, and Psychology), and to develop a more comprehensive list of linking adverbials. 5.1. Research aims and research questions The aims of this research were, within the corpus, to: (1) build up a list of linking adverbials in the four target categories; (2) investigate the frequency of all linking adverbials in the target categories; (3) investigate disciplinary variation; (4) investigate function. The following questions are directly addressed: (1) How frequently do RA authors use linking adverbials across a range of disciplines? Are there any interdisciplinary differences? (2) What linking adverbials do RA authors use across a range of LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 17 01 IBERICA 20.qxp 14/9/10 18:01 Página 17 disciplines? Are there any interdisciplinary differences? (3) How do linking adverbials function across a range of disciplines? Are there any interdisciplinary differences? 5.2. The RA corpus The corpus was 320 published RAs, forty from each discipline. Table 2 shows the length of disciplinary corpora. The eight disciplines were selected because they represent a range of subjects and also have large numbers of research writers around the world. This increases the usefulness of this research regarding recommendations for teaching ESP. Four leading refereed journals were selected from each discipline (see Appendix). The academic departments representing the eight disciplines were visited, and two sources from each were asked to name principal journals from their field. Ten RAs from 2007/2008 were randomly chosen from each journal by giving each a number and drawing numbers from a box. Only empirical data- driven RAs with the Introduction-Method-Results-Discussion format were chosen. Essays and discussions were excluded, and only one RA by any one writer was included in the corpus – that is, no writer’s name appears twice. No attempt was made to choose native-speaker authors, which in any case cannot be identified by merely looking at their name or the name of the institution where they work. It is suggested that the disciplinary corpora are sufficiently representative because of their size and because of the use of discipline sources to choose journals. MATTHEW PEACOCK Ibérica 20 (2010): 9-3418 LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): …-… 9 DISCIPLINE Number of RAs Total Word Length Science Chemistry 40 137,241 Computer Science 40 286,171 Materials Science 40 160,222 Neuroscience 40 243,057 Non-science Economics 40 292,488 Language and Linguistics 40 249,854 Management 40 285,825 Psychology 40 306,184 FOUR NON-SCIENCES 160 1,134,351 FOUR SCIENCES 160 826,691 ALL DISCIPLINES 320 1,961,042 Table 2. Lengths of disciplinary corpora. The eight disciplines were selected because they represent a range of subjects and also have large numbers of research writers around the world. This increases the usefulness of this research regarding recommendations for teaching ESP. Four leading refereed journals were selected from each discipline (see Appendix). The academic departments representing the eight disciplines were visited, and two sources from each were asked to name principal journals from their field. Ten RAs from 2007/2008 were randomly chosen from each journal by giving each a number and drawing numbers from a box. Only empirical data-driven RAs with the Introduction-Method-Results-Discussion format were chosen. Essays and discussions were excluded, and only one RA by any one writer was included in the corpus – that is, no writer’s name appears twice. No attempt was made to choose native-speaker authors, which in any case cannot be identified by merely looking at their name or the name of the institution where they work. It is suggested that the disciplinary corpora are sufficiently representative because of their size and because of the use of discipline sources to choose journals. 5.3. Investigating the corpus Analysis was done in the following steps, using the Concord and Contexts functions of WordSmith Tools 4.0 (Scott, 2004). Explanation of certain steps follows: STEP 1. A list of 46 linking adverbials was constructed from Biber et al. 1999, other grammars, and a thesaurus. This is more extensive than previously published lists. STEP 2. A preliminary examination of the corpus was conducted to check the function of all 46 linking adverbials, to see whether they do in fact always function as linking adverbials or not. This was done by individually checking a large number of occurrences of each in each discipline corpus 01 IBERICA 20.qxp 14/9/10 18:01 Página 18 5.3. Investigating the corpus Analysis was done in the following steps, using the Concord and Contexts functions of WordSmith Tools 4.0 (Scott, 2004). Explanation of certain steps follows: STEP 1. A list of 46 linking adverbials was constructed from Biber et al. 1999, other grammars, and a thesaurus. This is more extensive than previously published lists. STEP 2. A preliminary examination of the corpus was conducted to check the function of all 46 linking adverbials, to see whether they do in fact always function as linking adverbials or not. This was done by individually checking a large number of occurrences of each in each discipline corpus and evaluating function by reading the relevant sentence and surrounding sentences. At this stage it was found that eight, in addition to the dual- function items “though” and “so” discussed in the introduction, do not always function as linking adverbials: i. “rather” (exclude e.g. “rather quiet”, “rather tired”) ii. “yet” (exclude e.g. “we have not eaten yet”, “it is not yet clear”) iii. “alternatively” (exclude e.g. “content and language were alternatively the central focus”) iv. “similarly” (functions as a linking adverbial only in a sentence initial position. An example as non-sentence initial is “men and women behave similarly”) v. “further” (functions as a linking adverbial only when followed by a comma) vi. “in the same way” (functions only in a sentence initial position. An example as a non-sentence initial is “children respond in the same way”) vii. “that is” (functions as a linking adverbial only when followed by a comma) viii. “specifically” (functions only in a sentence initial position. An example as non-sentence initial is “the present study set out to specifically investigate”). STEP 3. The frequency of all linking adverbials was checked, along with disciplinary variation. All cases of “though” and “so” plus the above eight items not functioning as such were naturally excluded from the LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 19 01 IBERICA 20.qxp 14/9/10 18:01 Página 19 count: this required manual examination of every occurrence of each using the Concord function. Conjunctions (e.g. “although”, “while”, and “whereas”) were also excluded, because they are not linking adverbials (Biber et al., 1999; Carter & McCarthy, 2006). STEP 4. The function of every occurrence of all linking adverbials was individually checked by reading the relevant sentence and surrounding sentences. STEP 5. The frequency of all linking adverbials was examined within individual journals, to check if any were used significantly more frequently by certain journals. STEP 6. Statistical significance was set at p<.05 and tested with the chi- squared test within the log-likelihood calculator. Regarding steps 2 and 3, the corpus was split into disciplinary corpora at times to check disciplinary variation. Regarding steps 2, 3, and 4, “function” means “operates” or “acts”. Individual manual checking of the function of every occurrence is vital3. Two evaluators were involved in step (4): this writer and a local university lecturer. The second coder independently evaluated the function of every occurrence in order to measure inter-rater agreement. This writer reassessed the function of every occurrence after one month in order to measure intra- rater agreement. Inter-rater agreement was 97%, rising to 100% after discussions. Intra-rater agreement was 99%. 6. Results This section will start with the results for the three research questions. A total of 23,544 functioning linking adverbials were found in the whole corpus. The frequency of use was 12,006 per million words, or 74 per RA, though the number was drastically lower in Chemistry (24 per RA) and much lower in Materials Science (36 per RA). The number per RA for the other disciplines was Economics 89, Language and Linguistics 84, Management 97, Psychology 107, Computer Science 84, and Neuroscience 68. The authors of all 320 RAs used them. Frequency across all four semantic categories, and disciplinary differences, are shown in Tables 3, 4, 5, and 6 per million words (pmw). The forms are all in frequency order, with the most common first. Table 3 does not show all the contrast/concession forms, only those with a frequency of 100 pmw or above. MATTHEW PEACOCK Ibérica 20 (2010): 9-3420 01 IBERICA 20.qxp 14/9/10 18:01 Página 20 LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 21 MATTHEW PEACOCK Ibérica 20 (2010): …-…12 ITEM All 4 non- sciences ECON. LANG. MANAG. PSYCH. however 1380* 1401 1416 1308 1399 rather 456* 291* 560* 545 444 though 220* 164 256 217 248 in contrast/by contrast 205 182 176 133* 317* instead 184 202 160 150 219 on the other hand 174* 202 168 136 186 nevertheless/nonetheless/ notwithstanding 164* 96* 228* 119 219* otherwise 122* 295* 56* 80 49* TOTAL 3172* 3027 3292 2941* 3428* ITEM All 4 sciences CHEM. COMPU. MATS. NEURO. however 1110 905* 979 1219 1313* rather 305 161* 427* 138* 354 though 139 80 171 144 132 in contrast/by contrast 219 109* 140* 231 366* instead 170 212* 192 69* 189 on the other hand 111 175 112 144 53* nevertheless/nonetheless/ notwithstanding 76 7* 91 44 119 otherwise 64 51 112* 25* 41 TOTAL 2426 1832* 2434 2206 2905* * = statistically significant difference Table 3. Contrast/concession linking adverbials: Discipline differences. Frequency pmw. LINKING ADVERBIALS IN RESEARCH ARTICLES ITEM All 4 non- sciences ECON. LANG. MANAG. PSYCH. thus 914* 839 796 675* 1304* therefore 696* 801 488* 930* 546* so 412 469 568* 381 258* hence 292* 586* 116* 294 154* as a result 108 127 112 108 85 consequently 101* 82 88 143 92 thereby 72* 86 52 105 46 accordingly 71* 65 20* 84 105 for this reason 17 17 36 14 3* TOTAL 2682* 2921* 2204* 2545 2441* ITEM All 4 sciences CHEM. COMPU. MATS. NEURO. thus 644 489* 643 488* 840* therefore 613 533 612 569 691 so 403 358 174* 46* 64* hence 169 131 234* 219 82* consequently 69 117 73 75 33* thereby 41 29 45 44 41 accordingly 37 7* 49 38 41 for this reason 16 15 28 19 0* as a result 117 80 154 200* 41* TOTAL 2110 1723* 2353* 1856* 1951 * = statistically significant difference Table 4. Result/inference linking adverbials: Discipline differences. Frequency pmw. 01 IBERICA 20.qxp 14/9/10 18:01 Página 21 MATTHEW PEACOCK Ibérica 20 (2010): 9-3422 MATTHEW PEACOCK Ibérica 20 (2010): …-…14 ITEM All 4 non- sciences ECON. LANG. MANAG. PSYCH. also 2196* 1908* 2588* 2385 1974* as well 474* 291* 632* 605* 399 in addition 387* 387 392 353 415 moreover 196 315* 72* 175 203 furthermore 183 178 112* 231 199 Similarly 134* 92 140 161 144 further, 64* 51 72 84 52 additionally 59 45 60 59 72 likewise 37 21 48 28 52 besides 26 14 8 56* 26 In the same way 2 0 0 3 3 TOTAL 3758* 3301* 4124* 4140* 3539 ITEM All 4 sciences CHEM. COMPU. MATS. NEURO. also 1846 1263* 2105* 1856 1872 as well 368 299 448 338 333 in addition 326 197* 378 281 370 moreover 160 36* 150 231 193 furthermore 179 146 189 100* 239 Similarly 81 44 108 75 74 further, 40 22 70 13 33 additionally 53 66 42 69 49 likewise 30 58 31 19 21 besides 27 7 21 56 25 In the same way 5 15 0 0 8 TOTAL 3115 2153* 3542* 3038 3218 * = statistically significant difference Table 5. Addition linking adverbials: Discipline differences. Frequency pmw. LINKING ADVERBIALS IN RESEARCH ARTICLES ITEM All 4 non- sciences ECON. LANG. MANAG. PSYCH. e.g. 905* 473* 700* 706* 1670* for example/for instance 862* 795 952 934 784 such as 840* 507* 1152* 1276* 493* i.e./that is, 780* 736 688 448* 1209* in other words 136* 75* 188 133 154 Specifically 84* 96 20* 112 98 namely 68 62 100 73 42 TOTAL 3674* 2743* 3800 3682 4451* ITEM All 4 sciences CHEM. COMPU. MATS. NEURO. e.g. 711 88 863* 300* 1156* for example/for instance 504 175 899* 194* 432* such as 661 628 874* 625 457* i.e./that is, 561 285* 573 519 733* in other words 54 22 98* 13* 49 Specifically 52 15 52 50 74 namely 57 80 31 63 70 TOTAL 2601 1292* 3392* 1763* 2971* * = statistically significant difference Table 6. Apposition linking adverbials: Discipline differences. Frequency pmw. 6.1. Frequency – semantic categories Three categories, contrast/concession, addition, and apposition, were found to be more common than previously thought. Frequency per million words over all eight disciplines was contrast/concession 2857, result/inference 2441, addition 3487, and apposition 3221 whereas Biber et al. (1999) report 1200, 3000, 1000, and 1800 respectively – though of course their corpus was 50% book extracts and 50% RAs. These four tables reveal some striking interdisciplinary variations in linking adverbial frequency. The first clear difference is that the sciences 01 IBERICA 20.qxp 14/9/10 18:01 Página 22 6.1. Frequency – semantic categories Three categories, contrast/concession, addition, and apposition, were found to be more common than previously thought. Frequency per million words over all eight disciplines was contrast/concession 2857, result/inference 2441, addition 3487, and apposition 3221 whereas Biber et al. (1999) report 1200, 3000, 1000, and 1800 respectively – though of course their corpus was 50% book extracts and 50% RAs. These four tables reveal some striking interdisciplinary variations in linking adverbial frequency. The first clear difference is that the sciences use significantly fewer linking adverbials than the non-sciences in all four semantic categories. However, this science/non- science difference varied by discipline: Chemistry used all four types significantly less often. Materials Science used two types significantly less often – result/inference and apposition. Computer Science used three types significantly more often than the other sciences – result/inference, addition, and apposition; Neuroscience used contrast/concession and apposition significantly more often. Other variations may be seen among the non- science disciplines. 6.2. Frequency – individual forms The tables show the most common forms in each category. The first five in contrast/concession were “however”, “rather”, “though”, “in contrast/by contrast”, and “instead”. The most common in the other three categories were as follows: result/inference “thus”, “therefore”, “so”, and “hence”; addition, “also”, “as well” and “in addition”; apposition “e.g.”, “for example/for instance” and “such as”. No significant difference in the frequency of use of any linking adverbial was found within any one journal. Regarding contrast/concession, Table 3 shows a number of significant individual discipline differences in the use of individual forms. It also shows that the sciences used a narrower range of forms than did non-science authors: almost all forms in the table were used significantly less frequently by the sciences, except for the middle two, “in contrast/by contrast” and “instead”. Science authors favoured these two forms. Regarding apposition (Table 6), after careful examination of two pairs of forms across the entire corpus, “for example/for instance” and “i.e./that is”, it is concluded that they are interchangeable. LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 23 01 IBERICA 20.qxp 14/9/10 18:01 Página 23 6.3. Function Linking adverbials functioned only partly as predicted by Biber et al. (1999). Results from the four categories will now be presented, with numbered representative examples from the corpus: 6.3.1. Contrast/concession Biber et al. (1999) say they indicate alternatives. While this was found to be sometimes true in this corpus of RAs, they are also sometimes associated with making claims, and it is suggested that the latter function may be more important in RAs. First, here is an example denoting alternatives: (1) David could have used the touch handshape to represent grasping objects <1 in. in diameter and the small handshape to represent grasping objects between 1 and 2 in. in diameter. However, he did not use the forms in this way (Psychology). Second, here are two examples where the linking adverbial is associated with making claims. Example (3) uses two contrast/concession forms in one sentence. The second form marks the claim: (2) The drivers of value, which can be thought from the perspective of how a buyer actually evaluates a purchase situation, has been explicated to some degree in the purchasing management literature (…) However, what is missing is the interaction of the dyad and the preconditions (Management). (3) While the chronotopic approach is not intended to be used as a stand alone heuristic, but instead is intended to be used in conjunction with other teaching approaches, it is, nevertheless, an important step in coming to better understand what effects semantic and syntactic choices produce (Language and Linguistics). 6.3.2. Result/inference It is suggested that Biber et al.’s (1999) assertion that these linking adverbials mark results or consequences is correct. An example follows: (4) The criterion measure, B, was positively skewed and thus log transformed values were used for all analyses (Neuroscience). MATTHEW PEACOCK Ibérica 20 (2010): 9-3424 01 IBERICA 20.qxp 14/9/10 18:01 Página 24 6.3.3. Addition Biber et al. (1999) say that these show that the next unit is additional. While this was found to be often true in the corpus, it was noted that the very common form “also” often appears in chains. Additionally, “also” was often associated with making claims, and the latter function is assessed as being very important. First, here is an example showing the following unit is additional: (5) Also, while the value of the diversity of GCAE members is recognised, the difficulty of having a wide range of linguistic competence is also acknowledged (Language and Linguistics). Second, here are two examples of “also/as well” associated with claims. The forms also function to show that the next unit is additional, but it is suggested that to these authors, the “claim” function is particularly important. Example (7) contains three claims in one sentence: (6) Thus, we expect that the two theories would also be useful to predict the acceptance of SDA (Computer Science). (7) As well as providing a visual representation of each of the case companies, the framework also provides the means of comparing one cybermediary with another. The framework also offers a way of comparing any changes to the roles and relationships of a cybermediary over time (Management). 6.3.4. Apposition It is proposed that Biber et al.’s (1999) contention that these forms show the following text is an example or reformulation is correct. However, regarding the first of these functions – showing that the following text is an example – linking adverbials in this category had the additional function of helping authors to make claims. This can be seen in example (8). Example (9) shows “reformulation”: (8) The SV-mix approach also provides a useful tool for investigating the implications of some interesting hypothetical scenarios. For example, suppose that one wishes to think about the implications of a peso-type situation involving a low-probability, high-impact event (Economics). (9) An alternative explanation for our finding differences by behavior type may be that our stimulus behaviors differed in potency. In other words, perhaps our targets’ smiles were quite distinct (Psychology). LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 25 01 IBERICA 20.qxp 14/9/10 18:01 Página 25 Linking adverbials were found to be sometimes clustered together in long and complex sequences. These appear to be an important method of aiding in the strengthening of claims in RAs. Examples (3), (5), (6), (7) and (8) show this tendency. Another example follows: (10) Rather than thinking in terms of company promotions directly to consumers, for example, the team conceived joint advertising and promotions processes with the retailers, such as jointly sponsored mailings to retailer loyalty card holders. The team also realized that it had no product review procedure (Management). 7. Discussion and conclusions Linking adverbials were found to be more common than previously thought, though the corpora were of course not identical. Table 7 compares the present results with Biber et al. (1999): Semantic Category Biber et al. 1999 – LSWE academic prose Table 7 shows that in this study three categories (contrast/concession, addition and apposition) were much more common in RAs than in Biber et al.’s (1999) academic prose corpus (book extracts plus RAs). Result/inference was less common. These differences are striking. Another conspicuous difference is the proportion of three out of four semantic categories as a percentage of the whole. Contrast/concession made up around 24%, result/inference 20%, addition 29%, and apposition 25-28% (in this respect, the science and non-science disciplines were remarkably MATTHEW PEACOCK Ibérica 20 (2010): 9-3426 MATTHEW PEACOCK Ibérica 20 (2010): …-…18 (10) Rather than thinking in terms of company promotions directly to consumers, for example, the team conceived joint advertising and promotions processes with the retailers, such as jointly sponsored mailings to retailer loyalty card holders. The team also realized that it had no product review procedure (Management). 7. Discussion and conclusions Linking adverbials were found to be more common than previously thought, though the corpora were of course not identical. Table 7 compares the present results with Biber et al. (1999): Biber et al. 1999 – LSWE academic prose Present Study Frequency per million words Percentage of Total Semantic Category Frequency per million words Percentage of Total Non- science Science Non- science Science Contrast/ concession 1200 17 3172 2426 24 24 Result/ inference 3000 43 2682 2110 20 21 Addition 1000 14 3758 3115 28 30 Apposition 1800 26 3674 2601 28 25 TOTAL 7000 100 15055 11935 100 100 Table 7. Comparison of findings on four semantic categories: Biber et al. (1999) and present study. Table 7 shows that in this study three categories (contrast/concession, addition and apposition) were much more common in RAs than in Biber et al.’s (1999) academic prose corpus (book extracts plus RAs). Result/inference was less common. These differences are striking. Another conspicuous difference is the proportion of three out of four semantic categories as a percentage of the whole. Contrast/concession made up around 24%, result/inference 20%, addition 29%, and apposition 25-28% (in this respect, the science and non-science disciplines were remarkably similar) whereas Biber et al. (1999) report 17%, 43%, 14%, and 26% respectively. One cause of these category results is that the present search covered a greater number of linking adverbials than Biber et al. (1999), who do not list very many individual forms. An example of this can be seen in Table 1. Biber et al. (1999) report the frequency of the contrast/concession form “however” as 1100 pmw, and the whole contrast/concession category as 1200 pmw: “however” makes up 92%. Yet in this study, the frequency of “however” over all eight disciplines was 1266 pmw, making up only 44% of the whole category. On top of this the 01 IBERICA 20.qxp 14/9/10 18:01 Página 26 similar) whereas Biber et al. (1999) report 17%, 43%, 14%, and 26% respectively. One cause of these category results is that the present search covered a greater number of linking adverbials than Biber et al. (1999), who do not list very many individual forms. An example of this can be seen in Table 1. Biber et al. (1999) report the frequency of the contrast/concession form “however” as 1100 pmw, and the whole contrast/concession category as 1200 pmw: “however” makes up 92%. Yet in this study, the frequency of “however” over all eight disciplines was 1266 pmw, making up only 44% of the whole category. On top of this the frequency results for other individual forms were not much different from those reported in Table 1, yet the category frequencies are higher. Another important factor probably associated with the present results is the somewhat different corpora – Biber et al. (1999) used academic prose, book extracts plus RAs, around 50% each, and this factor must not be overlooked. Turning to the meaning of the present results, they indicate that contrast/concession, addition, and apposition linking adverbials are more common and therefore more important in RAs than previously thought, if readers of Biber et al. (1999) assumed that their results applied to RAs. And the higher frequency of the contrast/concession category makes it appear particularly important. It contains a large number of linguistic forms (see Table 3) available to and used by authors, and the functions that are reported here, such as assisting authors to construct claims, make it perhaps very valuable for them. The same is also true for addition and apposition linking adverbials, considering their higher frequency and extra functions. And regarding the function of linking adverbials across all four categories, it is suggested that this study adds to Biber et al.’s (1999) descriptions. Biber et al. (1999) said in their broad description that linking adverbials signal relationships between two units of discourse and create textual cohesion (they only mention presenting and supporting claims in connection with the result/inference category), and in their more detailed description that they indicate alternatives; show results and consequences; or show the next unit is additional, an example, or a reformulation. All this is true but it is argued here that in the RA contrast/concession forms are also sometimes used to help authors make claims; addition forms often assist in introducing claims; and apposition forms help writers strengthen claims. Additionally, linking adverbials were found to be sometimes clustered together in long and complex sequences, which appear to be another important method of LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 27 01 IBERICA 20.qxp 14/9/10 18:01 Página 27 supporting claims in RAs. It was noted in the introduction that it is very important for RA authors to persuade readers of the authenticity and validity of their claims, convince peers, and establish credibility, and therefore it is proposed that aiding writers with these functions is an important task. Regarding individual disciplines, the sciences use significantly fewer linking adverbials than the non-sciences in all four categories. But a closer look at the results shows that this varied sharply by discipline, and that the major difference was in two of the four sciences – Chemistry and Materials Science. A much closer examination of Chemistry and Materials Science RAs was then undertaken to try to understand some of the reasons for this much lower rate of occurrence. It was observed that authors tended to present and to develop claims in a different way – they used less argument. They described their research justifications, methods, results and conclusions in a much more narrative and descriptive style: they seemed merely to describe the steps they took, and their findings, one by one, and let readers work out their claims. Presumably the aim is to show the order of events or rather, this order is sufficient for readers, who perhaps do not need to be explicitly told the connections between facts, arguments, and claims. These authors did less restating, reformulating, exemplification, and less linking units of discourse together. Computer Science and Neuroscience RAs, on the other hand, resembled the non-science disciplines much more. Some typical and representative examples of these factors from Chemistry and Materials Science from the corpus will now be shown. The first is from a typical Chemistry “justify our research” section: (11) All current materials of this type use simple inorganic fluorides such as NaYF4 as host structures for luminescent lanthanide cations (…) With this in mind, we have recently initiated an exploratory study of Group III element fluoride hydrothermal chemistry, using organic amines as structure-directing agents, in order to open up a new area of structural solid-state chemistry. The next example is from a typical Chemistry conclusion section: (12) Here, it has been shown that MF-ICA is capable of providing good estimates of the elution patterns containing background, mass spectral sources and the area of the eight analytes in four clusters despite their heavily overlapped signals. The reasonable results would be obtained by selecting the proper parameters, such as number of sources, non- negative mixing prior and source prior values. MATTHEW PEACOCK Ibérica 20 (2010): 9-3428 01 IBERICA 20.qxp 14/9/10 18:01 Página 28 The next example is from a typical Materials Science “introduce our research” section: (13) Spraycasting followed by hot working in this study is designed to produce an intermediate grain size in the range 1-5 um without the need for severe deformation. The present paper describes an investigation of the microstructure of a spray cast Al–5.31Mg–1.15Li–0.28Zr alloy and its microstructural evolution during hot compression as a function of temperature. Example (14) is from a typical Materials Science conclusion: (14) A high quality Al–5.31Mg–1.15Li–0.28Zr billet with an initial grain size of -10UM has been produced by spraycasting. The Zr micro- distribution was non-uniform, with denuded regions adjacent to coarse L12Al3Zr particles at grain boundaries. After consolidation and dispersoid precipitation heat treatment at 400 deg. C for 4 h the grain size coarsened. A different style is apparent in the following Computer Science and Neuroscience extracts. Example (15) is from a Computer Science discussion section: (15) The main advantages of CVA compared to HMM are that CVA is easy to implement and does not require such complex operations as HMM. If one wants to avoid the burden of calculation of probability density functions while building a classifier, CVA is a reasonable choice. But in the CVA method, a whole utterance is treated as a single vector instead of a sequence of independent vectors. Therefore, the number of parameters in CVA is larger. The last extract is from a typical Neuroscience conclusion: (16) phMRI is an exciting new non-invasive approach to study receptor function in healthy, diseased and treated subjects. It is the only non- invasive imaging approach that allows mapping of pharmacological interactions in vivo with a high spatial and temporal resolution (Borsook et al., 2006). Developing imaging approaches that will allow in vivo monitoring of cell transplants is essential to advance this therapeutic approach efficiently into clinical practice (Modo et al., 2004). The use of phMRI to assess disease and treatment will surely provide exciting novel insights. LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 29 01 IBERICA 20.qxp 14/9/10 18:01 Página 29 One method or further examining and helping to explain the much lower rate of occurrence in these two disciplines would be to do a text analysis of contrasting examples with the same discourse function or sub-function in different RA sections, for example introduction-discussion-conclusion, and investigate to what extent authors achieve these functions using linking adverbials. However, this particular area of extended research is beyond the scope of the present study. Some of the many significant discipline differences with individual forms may be explained by a presumed interchangeability of some of the forms. For example, Economics used “rather” far less often and “otherwise” far more often. Yet for two disciplines, this does not explain the variation: Management and Materials Science authors did not use any contrast/concession forms more often. 7.1. Pedagogical implications Linking adverbials seem to be important, and these results have implications for teaching ESP and particularly for the teaching of research writing, especially dissertations and research papers. It is suggested that competence in research writing includes a developed knowledge of linking adverbials, i.e. when, where and how to use them. This knowledge varies across disciplines and this implies that an awareness of discipline variations is necessary for teachers of research writing. Certainly students must be told of the importance of context in using linking adverbials in research writing, and be told that awareness of their use is necessary4. 7.2. Further research More complete lists of linking adverbials need to be developed, perhaps combining lists used in previous research, forms found in dictionaries and a thesaurus, and forms found inside RAs. The latter method may be especially useful. Other questions to research are: How frequent are linking adverbials in other disciplines, and how are they used? What other functions do they have across disciplines? How and when are they acquired by research writers? Exactly how do they function when they cluster together in long and complex sequences? How do Chemistry and Materials Science authors achieve certain functions? This study shows the importance of linking adverbials in RAs in the sciences MATTHEW PEACOCK Ibérica 20 (2010): 9-3430 01 IBERICA 20.qxp 14/9/10 18:01 Página 30 versus non-sciences, and also in individual disciplines. It is proposed that linking adverbials are more important in RAs as signalling and cohesive devices, and for helping authors manoeuvre more effectively to make and/or to strengthen claims, than previously thought. Also, different disciplines achieve this in significantly different ways, confirming the importance of discipline variation when researching their use, and adding to knowledge of ESP. It is hoped this study helps us better understand scientific expression and the RA. [Paper received December 2010] [Revised paper accepted April 2010] References LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 31 Berkenkotter, C. & T.N. Huckin (1995). Genre Knowledge in Disciplinary Communication: Cognition/Culture/Power. Hillsdale, NJ: Lawrence Erlbaum. Bhatia, V.K. (2000). “Genres in conflict” in A. Trosborg (ed.), 147-161. Biber, D. (2006). University Language: A Corpus- Based Study of Spoken and Written Registers. Amsterdam: John Benjamins. Biber, D., S. Johansson, G. Leech, S. Conrad & E. Finegan (1999). Longman Grammar of Spoken and Written English. Harlow: Pearson Education. Bolton, K., G. Nelson & J. Hung (2003). “A corpus- based study of connectors in student writing: research from the International Corpus of English in Hong Kong (ICE-HK)”. International Journal of Corpus Linguistics 7: 165-182. Carter, R. & M. McCarthy (2006). Cambridge Grammar of English. Cambridge: Cambridge University Press. Charles, M. (2007). Using a Corpus to Explore Argument in Academic Discourse: How Far Can We Go? Paper presented at Corpus Linguistics 2007, University of Birmingham, UK, 27-30 July. Chen, C. W-Y. (2006). “The use of conjunctive adverbials in the academic papers of advanced Taiwanese EFL learners”. International Journal of Corpus Linguistics 11: 113-130. Conrad, S.M. (1999). “The importance of corpus- based research for language teachers”. System 27: 1-18. Conrad, S. (2000). “Will corpus linguistics revolutionize grammar teaching in the 21st century?” TESOL Quarterly 34: 548-560. Hoey, M. (2001). Textual Interaction. London: Routledge. Hopkins, A. & T. Dudley-Evans (1988). “A genre- based investigation of the discussion sections in articles and dissertations”. English for Specific Purposes 7: 113-121. Hunston, S. (1994). “Evaluation and organization in a sample of written academic discourse” in M. Coulthard (ed.), Advances in Written Text Analysis, 191-218. London: Routledge. Hunston, S. (2001). “Colligation, lexis, pattern, and text” in M. Scott & G. Thompson (eds.), Patterns of Text, 13-33. Amsterdam: John Benjamins. Hyland, K. (1996). “Talking to the academy: forms of hedging in science research articles”. Written Communication 13: 251-281. Hyland, K. (1997). “Scientific claims and community values: articulating an academic culture”. Language and Communication 17: 19-31. Hyland, K. (1998). Hedging in Scientific Research Articles. Amsterdam: John Benjamins. Hyland, K. (2000). Disciplinary Discourses: Social Interactions in Academic Writing. Harlow, Essex: Longman. Liu, D. (2008). “Linking adverbials: an across- register corpus study and its implications”. International Journal of Corpus Linguistics 13: 491-518. Moreno, A. I. (1997). “Genre constraints across 01 IBERICA 20.qxp 14/9/10 18:01 Página 31 Matthew Peacock teaches in the Department of English at the City University of Hong Kong. His research interests include English for Specific Purposes, corpus analysis, research writing, genre analysis, and TEFL methodology. He co-edited (with John Flowerdew, 2001) a collection from Cambridge University Press, Research Perspectives on English for Academic Purposes. NOTES 1. Liu (2008) offers a useful review of the literature on the two types of linking devices. 2. See Silver (2003) for further research on stance adverbials. 3. Many authors stress the importance of doing this: for example frequency is obtained from statistical analysis but context is vital in understanding function (Tognini-Bonelli, 2004), and a “microscopic study” must be carried out before categorisation can be done (Williams, 2002: 60). 4. See Bolton, Nelson and Hung (2003) and Chen (2006) for further ideas on teaching. MATTHEW PEACOCK Ibérica 20 (2010): 9-3432 languages: causal metatext in Spanish and English RAs”. English for Specific Purposes 16: 161-179. Paltridge, B. (1993). “Writing up research: a systemic functional perspective”. System 21: 175- 192. Scott, M. (2004). WordSmith Tools (Version 4). Oxford: Oxford University Press. Shaw, P. (2000). “Towards classifying the arguments in research genres” in A. Trosborg (ed.), 41-56. Silver, M. (2003). “The stance of stance: a critical look at ways stance is expressed and modeled in academic discourse”. Journal of English for Academic Purposes 2: 359-374. Swales, J.M. (1990). Genre Analysis: English in Academic and Research Settings. Cambridge: Cambridge University Press. Swales, J.M. (2004). Research Genres: Explorations and Applications. Cambridge: Cambridge University Press. Tognini-Bonelli, E. (2004). “Working with corpora: Issues and insights” in C. Coffin, A. Hewings & K. O’Halloran (eds.), Applying English Grammar: Functional and Corpus Approaches, 11-24. London: Arnold. Trosborg, A. (ed.) (2000). Analysing Professional Genres. Amsterdam: John Benjamins. Williams, G.C. (1998). “Collocational networks: interlocking patterns of lexis in a corpus of plant biology research articles”. International Journal of Corpus Linguistics 3: 151-171. Williams, G. (2002). “In search of representativity in specialised corpora: categorisation through collocation”. International Journal of Corpus Linguistics 7: 43-64. 01 IBERICA 20.qxp 14/9/10 18:01 Página 32 Appendix: Journals in the corpus LINKING ADVERBIALS IN RESEARCH ARTICLES Ibérica 20 (2010): 9-34 33 MATTHEW PEACOCK Ibérica 20 (2010): …-…24 Appendix: Journals in the corpus Science Non-science Chemistry • Analytica Chimica Acta • Inorganica Chimica Acta • Journal of Organometallic Chemistry • Journal of Solid State Chemistry Economics • Economic Modelling • Journal of Economic Behavior and Organization • Journal of Economics and Business • Journal of Financial Economics Computer Science • Computers in Human Behavior • Computer Speech and Language • Information and Software Technology • International Journal of Human- Computer Studies Language and Linguistics • English for Specific Purposes • Journal of English for Academic Purposes • Journal of Second Language Writing • System Materials Science • Acta Materialia • Biomaterials • Corrosion Science • Polymer Management • Journal of Business Venturing • Journal of International Management • Industrial Marketing Management • International Journal of Information Management Neuroscience • Cognition • Brain and Cognition • Neuropsychologia • Neuroscience Psychology • Acta Psychologica • Cognitive Psychology • Journal of Anxiety Disorders • Journal of Research in Personality 01 IBERICA 20.qxp 14/9/10 18:01 Página 33 01 IBERICA 20.qxp 14/9/10 18:01 Página 34