Harnessing Sources in the Humanities: A Corpus-based Investigation of Citation Practices in English Literary Studies

Integrating outside sources for rhetorical purposes is an essential element of academic writing; not only is citation a vehicle for writers to situate their research within a continuum of disciplinary knowledge, it is also a way to establish authorial ethos . Yet effective referencing can be problematic for academic writers, especially novices. Corpus-based research into science writing has provided valuable insight into how published scholars work with sources; however, the citational practices of research article writers in the humanities have remained largely unexplored. This paper analyses a 35-article corpus drawn from the field of English literary studies and reveals the distinctive citational practices of authors in that field. The results reported here problematize assumptions that standard writing practices exist across humanities disciplines or that extensive commonalities exist with social science writing. Important findings include that literary studies authors cite relatively less, and, when they do, they favour quotation over paraphrase and summary unlike writers in previously-examined fields. As well, their syntactic integration of references and reporting verbs differ significantly from other disciplines. The results reported here provide valuable support for discipline-specific graduate writing instruction and underline the need for further research into humanities writing practices.

monographs or journals (Knievel and Kellsey, 2005). Yet little attention has been paid to the ways in which those sources actually manifest themselves in scholarly texts.
In the discipline considered in the present research, English literary studies, investigations into the rhetorical, syntactic, and lexical aspects of using outside sources has been a relatively rare occurrence. In one of the few exceptions, Shaw (2001) compared a corpus of published literary studies articles to one comprised of undergraduate essays. Quantitatively focused largely on the usage frequency of key words, this research included some work with speech act verbs; however, its findings are of limited interest in the present context because every occurrence of a reporting verb was counted, regardless of whether they were being used for secondary source citation or not -i.e. a verb used to attribute an utterance to a literary character would be included. Reaching farther back, MacDonald's (1992) sentence-level investigation of four literary studies articles found that only 5% of grammatical subjects fell under the "research" category, that is, "references to scholars in the field" (p. 544). That finding led MacDonald to conjecture that "literary academics" accorded less significance to "co-operative disciplinary knowledge" when compared to psychology and history scholars (p. 547). More recently, Šinkūnienė (2017) used corpus analysis to compare the citation practices of English and Lithuanian research articles in literary studies and linguistics to the writings of Lithuanian undergraduate students. Using a relatively small corpus of ten literary studies articles, she found a preference for integral over non-integral forms and for quotation over paraphrase.
Scholarly motivations for citing previous research have been an area of investigation in literary studies as well. For example, Frost's (1979) overview of German literary criticism concluded that two prime reasons were to support one's own argument and to demonstrate that a range of opinion exists. Similarly, Tsay and Chiu (2014) examined citations in a range of humanities journals, including those from literature studies, and found that writers tended to use outside sources to provide factual The centrality of referencing sources in academic writing has, however, spurred a substantive body of research into student writers' citation practices (e.g. Charles, 2006;Davis, 2013;Hirvela & Du, 2013;Keck, 2014;Pecorari, 2006;Petric 2007;Schembri, 2009;Swales, 2014;Thompson and Ye, 1991;Yeh, 2009). A recent extensive survey (Cumming et al., 2016) that focused on research into how citation issues impact L2 learners bears testament to the level of interest EAP scholars have shown in the subject. Such research is invaluable as it allows educators insight into areas in which graduate writers need instruction; for example, Thompson and Tribble's (2001) study of doctoral theses revealed that students' range of citation practices were limited in multiple ways. In response to such shortcomings, they recommended that learners analyse exemplary texts in order to familiarize themselves with best writing practices in their field. Similarly, in a discussion of how to improve graduate writing skills, Schilb (2002) advocates that students "consider ways in which their own writing matches or departs from their discipline's main rhetoric" (p. 141). Undoubtedly, acquainting learners with patterns of citation employed by published authors in their field can be a potent tool to help graduate students achieve their writing goals. Yet writers in the humanities have had to date little corpus-based research in that area to draw upon in order to discern best practices.
While the limited research into citation in literary studies outlined above may well be useful to graduate writers looking for guidance on what to cite and why to cite, it sheds little light on issues surrounding how to cite. Hellqvist (2010) argues that further studies into the role of referencing in the humanities are needed to shed light on problems and possibilities in both informetric research and in knowledge organization. Studies of citing patterns in specific disciplines as well as more broad studies of scholarly communication within the humanities are needed, and quantitative methods (e.g., bibliometric) as well as qualitative methods can be used in this effort. (p. 316) Speaking specifically of English literary studies, Wilder (2012) maintains that "for a map of topoi and stasis practices that would be useful for students of literature and WID scholars, analysis of a more recent diverse corpus is needed" (p. 23). My research responds to both Hellqvist's and Wilder's challenges by using corpus-based analysis to examine how published authors in English literary studies incorporate the ideas and words of others into their writing. This field is of particular interest given the rhetorical exigencies Schilb (2002) identifies as follows: Graduate students writing about a literary text tend to face certain additional challenges. For one thing, they wonder how best to acknowledge previous analysts of the text …. how to appropriate past scholarship so as to make their own argument cogent. (p. 146) Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw 29 In a discipline that, like so many in the humanities, depends on new ideas being situated within ongoing discussions, recognizing how experts integrate their sources is essential.
In order to meet this disciplinary need and, at the same time, expand the limited research into humanities citation practices, this paper quantitatively analyses a corpus of published English literary studies research articles. The choice of methodology is apt; as Hyland (2015b) argues, "corpus approaches to academic writing provide insight into disciplinary practices which help explain the mechanisms by which knowledge is socially constructed through language" (p. 292). In all, 35 articles were analysed to gauge the following: the extent to which sources are used; how writers syntactically incorporate their sources; whether writers prefer to paraphrase or quote; if material is imported verbatim, what the norms are for quotation length; and, finally, what vocabulary is used to connect source authors to their ideas. A secondary objective of this research is to contextualize its findings within previous work, in particular Hyland (2004), in order to further two important discussions: first, whether generalizations are indeed possible regarding best writing practices in the humanities, and, second, whether assuming affinities between writing practices in the social sciences and the humanities under umbrella terms such as the "soft sciences" is potentially problematic.

Methodology
Thirty-five articles, drawn from three English literary studies journals (PMLA [the journal of the

Modern Language Association of America], English Literary History [ELH] and Studies in English
Literature [SEL]), comprised the corpus for this study. While monographs may be cited more often in the humanities (Knievel and Kelsey, 2005;Thompson, 2002), the research article was chosen as the target form as it is the most likely genre for aspiring academic writers to emulate as they attempt to establish themselves as scholars. For many graduate writers, converting a chapter from a thesis-inprogress into a journal article will provide their entrée into the publishing world. As well, journal articles, for a variety of reasons, appear to be gradually eclipsing monographs as the predominant scholarly venue for disseminating research (Hyland, 2015a).
The criteria for corpus inclusion were as follows. Following the lead of previous corpus research (e.g. Harwood, 2005;Hyland, 2004;Kuhi & Behnam, 2011), the source journals were selected via "informant nomination"; specifically, professors of English at a major Canadian university were asked to identify important generalist journals that published critical articles on a wide range of English literary periods, genres, and authors. Articles were then selected that critically engaged with a single Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw literary work, traditionally a dominant genre mode for writers in this field. Articles were restricted to this form to minimize variances in the data produced; as Crookes (1986) observes, article type has an impact on rhetorical organization. In this case, for example, research dealing comparatively with multiple primary literary texts (a less dominant mode in this field) and drawing evidence from such, could potentially skew the extent to which secondary sources are employed within the fixed word counts imposed by journals. Authorial traits, such as gender, publishing experience, and native speaker status were not considered in the selection process. A full list of the articles comprising the corpus may be found in Appendix A.
After assembling the corpus, each instance of citation was manually tagged using a custom-built program. Citations were defined as any in-text reference to a non-literary primary or secondary text or author. Included in the primary text category were historical, biographical, philosophical, and theoretical works, as well as diverse writing genres such as letters and book reviews. Secondary sources were defined as per Budd (1986): "those materials that report analysis, interpretation or background information based on the work of scholars and/or critics" (p. 193). Given that the focus of the present research was the integration of outside sources into the rhetorical flow constructed in the main body of the text, cited materials newly introduced in footnotes (which can serve a variety of discursive functions, some not directly related to the primary argument being advanced (Heinzkill, 1980) were excluded from consideration.
Each referencing instance was manually tagged as per the following parameters. The first distinction drawn was between integral and non-integral citation, the former being when a source is named in the main text, the latter when attributions are made either parenthetically or in a footnote or endnote. All integral citations were further broken down according to how the sources were syntactically incorporated into the text. Five categories were used to differentiate between methods of introducing named sources; three followed Hyland's tripartite taxonomy (2004): in the sentence subject position (e.g. "Knudson argues the impossibility of…"), in a non-subject position (e.g. "That impossibility led Sanchez to claim…"), and as a noun phrase/ possessive form (e.g. "Smith's view is that…"). Two further forms of source integration emerged in sufficient numbers during the tagging process to merit separate recognition as standard syntactic strategies in English literary studies: as an adjunct (e.g. "According to Zhang…") and in a mid-sentence attribution consisting of a verb and source without a conjunctive "that" or prepositional link joining the reporting verb to the independent clause (e.g. "Previous approaches," Corkin argues, "have been biased."). In Hyland (2004), these last two categories presumably fell under the "non-subject" heading. Once again, in Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw keeping with previous studies, "generalizations" were also tagged, i.e. when the same idea is attributed to a group of individual sources (e.g. "Many critics have argued that...") (Hyland, 2004).
Both integral and non-integral citations were tagged according to whether the source idea was imported via direct quotation or by paraphrase. Quotations were further categorized according to length using parameters established in Borg (2000): "extended" being quotations longer than 40 words, "brief' being less than 40 words, and "fragment" being those of five words or less.
In all instances of integral citation in which a verb connected a source with the imported material, the reporting verb was tagged and then assigned one of three designations according to its discoursal function. Previous researchers categorizing reporting verbs have used a variety of designatory terms similar to those used in the present study; for example, Charles (2006) used "argue," "think," and "show" to differentiate between types of reporting clauses. Hyland (2004), divided reporting verbs according to whether they denoted "research acts," "cognition acts," or "discourse acts," a breakdown which respectively followed Thompson and Ye's (1991) taxonomy of "research verbs," "mental verbs," and "textual verbs". I adopted a taxonomy following Hyland (2004, p.27) that breaks down as follows: "cognition verbs" being those imparting a stance to the source author vis-à-vis the imported idea (e.g. "argues," "believes," "suggests," "advocates"); "research verbs" being those denoting a research act or finding (e.g. "found," "examined," "concluded," "discovered"); and "discourse verbs" being those connoting ostensibly non-subjective discourse acts such as "comments," "discusses," "writes," and "notes." A list of all tags used to collate data for this research and samples of tagged text can be found in Appendix B.

Number of Citations
Overall, the 284,871-word corpus of 35 literary studies articles yielded 1,200 instances of citation, resulting in an average of 34.3 references per paper and an average of 4.2 per 1,000 words. As evident in Figure 1, we can already start to see quantitative deviations from the disciplines surveyed in Hyland (2004). On the basis of average number of citations per paper, literary studies papers would land in seventh place on his list, and, from this very first comparison, Hyland's (2004) "informal characterization" that "softer disciplines tend to employ more citations" (p. 24) is problematized as writers in literary studies cited far less often than those who authored his social science corpus (see Figure 1). The most interesting comparison, however, comes when we consider that study's sole Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw 32 humanities representative, philosophy, and its average of 85.2 citations per paper and 4.2 instances per 1,000 words -rates more than double those in the present corpus. While Hyland's smaller corpus (10 articles vs. 35) may have been skewed by outliers, this basic quantitative difference is of such size that its significance is self-evident. The vast difference between the results in literary studies and Hyland's (2004) "soft disciplines" provide a first indication that a generalized view of citation practices and patterns across the humanities and social sciences is not sustainable.

Integral vs. Non-integral Citation
As evidenced in Figure 2, authors in the English literary studies corpus chose integral over nonintegral citation by a wide margin, thereby providing some common ground with Hyland's (2004) philosophy corpus. Overall, in the present corpus, the name of the source appeared in the main text 814 times (67.8% of the total citation count) compared to the 386 citations (32.2%) when the source was relegated to a parenthetical reference or footnote. These results are roughly in agreement with Hyland's (2004) findings in his philosophy sampling (64.6% integral vs. 35.4% non-integral). The preference for integral forms in English literature studies was shown whether the material was imported via quotation or by paraphrase/summary. Applied Linguistics (Hyland, 2004) Sociology (Hyland, 2004) Marketing (Hyland, 2004) Philosophy (Hyland, 2004) English Literary Studies Average number of citations per 1,000 words Given the argumentative nature of most humanities writing, positioning one's work within existing discussions or debates is a common approach for establishing the topic under discussion. As debates are more easily envisaged as occurring between individuals, a preference for integral forms in the humanities may be a strategy to achieve that rhetorical end; however, Gray's (2015) survey of history journal articles, which found a 100% preference for footnoted or endnoted references, problematizes across-the-board assumptions regarding humanities citation practices. That study did not statistically break down citation format into integral or non-integral, noting only whether the "primary means of citation" (p. 65) was through in-text references or not. Nonetheless, the preponderance of integrated forms found in the current research aligns with the results Šinkūnienė (2017) gleaned from a smaller corpus of ten English literary criticism research articles. Figure 2 contrasts the preference for integral forms in literary studies and philosophy with Hyland's (2004) social science corpora, in which it seems that the research is foregrounded and the individual researcher's role is, to a certain extent, minimized by being named outside the running textual narrative. This distinction between humanities and social science writing further reinforces the difficulties of generalizing writing practices across the two research areas.

Use of Generalizations
The extent to which broad claims arising from the literature are used in literary studies brings us once more into agreement with Hyland's (2004) philosophy/humanities results. In the present 0 20 40 60 80 100 Applied Linguistics (Hyland, 2004) Sociology (Hyland, 2004) Marketing (Hyland, 2004) Philosophy (Hyland, 2004) English Literary Studies

Integral vs. non-integral citation
Non-integral citation (by percentage) Integral citation (by percentage) because they prefer integral forms, as Šinkūnienė suggests (p. 267), or do they use integral forms because they quote more? I would venture that the latter correlation seems more plausible.

Figure 3. Quotation vs. Summary/paraphrase/generalization
Given that we might have expected a measure of commonality between humanities fields, the most startling difference between Hyland's (2004) philosophy articles and the present corpus lies in whether the original material was quoted or paraphrased/summarized; specifically, we can see that the results from Hyland's (2004) humanities/philosophy texts are diametrically opposed to the literary RA results, with 89 instances of summary enumerated and only three of quotation in the ten philosophy articles. On this important aspect of citation practice, the humanities texts in Hyland (2004) showed a far greater affinity with social science writing with regard to privileging paraphrased forms.
Certain factors associated with literary studies may help to account for such a pronounced difference from other fields when it comes to preferring direct quotation. First, this discipline is unlike most in the way that successful argumentation often hinges on the sustained utilization of evidence derived from primary texts, i.e. the works of literature under discussion. Specifically, the proofs regularly relied upon are the words found in the poems, plays, novels and other literary forms themselves. These words, of course, cannot be paraphrased, since what is actually written is either the object of analysis or evidence for a claim; conceivably, this habituation and the necessity of using verbatim quotations affects how secondary sources are presented within critical texts. In short, perhaps quoting is second nature. An additional factor potentially at play is the "mistaken critic" Applied Linguistics (Hyland, 2004) Sociology (Hyland, 2004) Marketing (Hyland, 2004) Philosophy (Hyland, 2004) English Literary Studies Quotation vs. Summary/paraphrase/generalization topoi observed by Wilder (2005) in her corpus of literary criticism research articles. She argues that one rhetorical element that contributes to "knowledge building" in this discipline is "meticulously documented citations of critics and theorists whose work is being corrected or amended" (p. 112).
Much like the data generated in experimental research, the words of other scholars often constitute the "raw material" from which new knowledge is generated through a reactive process. Yet any reconfiguration of source material almost unavoidably carries a degree of subjectivity; as Hyland notes, the opportunity to shape imported material may constitute a writer's advantage that accounts for paraphrasing's preponderance (Hyland, 2004, p. 26). In the end, introducing text imported verbatim for debate or refutation may justifiably be viewed less skeptically than paraphrases.
Hyland's assertions regarding the advantages of paraphrasing supported his results, which overwhelmingly illustrated a preponderance of paraphrasing in the social sciences and his humanities field; yet the preference for paraphrase does not seem to be the case in literary studies, and we are compelled to make allowances for field-specific rhetorical and citational practices that violate what might be perceived to be academic writing norms in the "soft" disciplines.

Integral Citation Syntactic Forms
In an integral citation, the named source can assume a variety of syntactic roles. This study looked beyond Hyland's (2004) three categories of integration form ("subject," "non-subject," and "noun phrase"). While that tripartite division may have served well for a largely science-oriented corpus, further breaking down Hyland's "non-subject" designation produces a more nuanced picture of how sources are integrated in literary studies by allowing the enumeration of structures that merited separate recognition, based on frequency of use in the corpus (as discussed above) . The value of this further division of forms is evident as one of the two new designations ("adjunct agent") turned out to be the second-most prevalent structure in the literary studies corpus.
The reporting structure used most frequently in the corpus was "source in subject position," a form in which the sentence focus rests on the person who originated the new information, arguably appearances were tallied in the corpus, the results indicating that writers in this discipline favoured cognition (315 occurrences, 46% of the total) and discourse verbs (299 -43.6%) -almost in equal numbers. The third category, research verbs, trailed far behind with 71 instances (10.4%).
Once again, situating the present research alongside past investigations is instructive. Hyland's (2004) corpus analysis revealed that his three social science fields and lone humanities discipline (philosophy), overwhelmingly favoured discourse verbs over cognition ones in ratios ranging from 4:1 to nearly 10:1. The literary studies corpus evidence demonstrates an important difference in disciplinary citing behaviours, namely a higher usage of verbs that denote a subjective relationship between the original source and the reported material, i.e. cognition verbs. Figure 4 illustrates the marked difference between the literary studies corpus' ratios of reporting verb categories and Hyland's (2004) social science and humanities ratios. The distribution in literary studies, which skews away from "research" verbs, is perhaps explained by quantifiable "proofs" being hard to come by in this field, unlike in the sciences.
Accordingly, it was not surprising to find positional stances being attributed to a large percentage of sources, given that "truths" are constructed in this field largely through argumentation. Yet it remains perplexing how this intuitive truth about the humanities was not manifested in Hyland's philosophy corpus .   0  10  20  30  40  50  60  70  80  90  100 Marketing (Hyland, 2004) App. Linguistics (Hyland, 2004) Sociology (Hyland, 2004) Philosophy (Hyland, 2004) Literary Studies Types of reporting verbs used (by percentage)

Discourse Cognition Research
Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw 40 The ratio of imported material in the literary studies articles being relayed via discourse verbs, was more in line with Hyland's (2004) social science/humanities fields. One possible explanation for the extensive use of verbs such as "discuss" or "writes" in the two humanities fields under discussion is that the neutral conveyance of ideas may rhetorically remove some of the hedging and subjectivity associated with them, i.e. the reality of hard proofs being difficult to come by because people can really only "propose" or "suggest" things about, in the case of literary studies, the intentions of a text's author. By not foregrounding that something is "argued" or "contended," writers may be subtly attempting to add a patina of objectivity to previous work and thereby further the acceptance of their sources' contribution to the discussion. Of particular interest is that the number one reporting verb in the literary studies corpus-"note"-seems fairly distinctive to the field, at least on a quantitative level. It appears only once in Canadian Journal for Studies in Discourse and Writing/Rédactologie Volume 29, 2019 http://journals.sfu.ca/cjsdw Hyland's (2004) lists, placing fourth in verbs used in a social science field (Sociology) and not at all in the top-seven list drawn from Philosophy -his humanities field. In contrast, the verb most frequently used in Hyland's (2004) philosophy corpus, "say," was relatively shunned in English literary studies, appearing only seven times, registering an almost-negligible frequency of 1.02%.
The number two and three reporting verbs in the literary studies corpus, "argue" and "suggest," were perhaps more predictable and also ranked highly in Hyland's frequency lists for his social sciences and philosophy corpora. The fourth-place verb, "call," does not appear in Hyland's (2004) lists at all, and as we move down the literary studies list, there are further notable variations from other disciplines: to wit, the fifth-place verb in the present study, "observe," whose frequency is almost equal to that of "suggest," appears in Hyland (2004) only in biology, a life sciences discipline, as the sixth-most-used verb. One would suspect, however, that this verb is used differently in the two fields -i.e. as a discourse act in literary studies and as a research verb in biology; however, this remains unclear as the frequency lists in Hyland (2004) are not broken down into function categories. The sixth most commonly used verb in the literary studies corpus, "write," does not appear in Hyland's (2004) lists. The seventh-highest-frequency verb in the literary studies corpus, "describe," shows up in Hyland's list as a verb used in three of his "hard" sciences and one "soft" (sociology). In the end, Table 1 reveals that the reporting verbs on the literary studies frequency list bear little semblance to those found in Hyland's (2004) humanities field, philosophy. Comparing the top seven (the extent of Hyland's reported results) verbs in the two fields yields only a 28.5% overlap, reinforcing the impression that disciplinary discourse communities develop their own "language." the present study, the only detailed piece of research into humanities citation practices in published work (as opposed to in student and in L2 writing) has been Hyland's (2004) study of a corpus of philosophy articles. The profound differences between that study's results and the present investigation in areas such as quotation vs. paraphrasing and rates of citation further emphasizes that humanities writing proclivities are far from monolithic.
Based on the results reported here, grouping together the so-called "soft" disciplines, i.e. the humanities and social sciences, in order to generalize about writing practices seems equally problematic. For example, consider how Pecorari's (2006) claim, that "it was expected that the science and engineering writers would avoid explicit quotation, and that those in the humanities and social sciences would use it as a minority strategy, and this proved to be the case" (p. 20) is undermined by the preference shown by literary studies authors to directly quote their sources.
Further, on issues such as whether to favour integral or non-integral forms, the humanities and social sciences, based on disciplinary research to date, would appear to be at loggerheads; yet, while this would suggest a definable difference between the two research areas, two previous investigations of thesis writers in the social sciences, Thomson (2000) and Charles (2006), suggest otherwise, as they found integral forms to be the dominant mode in agricultural economics and political science, respectively. Whether published articles in those fields would yield similar results remains unknown, but, at the very least, a potential binary of humanities/integral versus social sciences/nonintegral is problematized by this prior research into doctoral candidates' writing practices.
Such disjunctures between disciplinary writing practices present significant problems when aspiring graduate writers encounter research suggesting, for example, that paraphrase and summary should be privileged over quotation when, as demonstrated above (3.4), research articles in one field demonstrably suggest otherwise. Academic writers who turn to Google looking for guidance on incorporating sources will find no shortage of university writing websites dispensing advice such as the following from Leeds University: "…Quotes should be using sparingly as over quoting can suggest a lack of understanding of the text you are referring to." ("Citing quotations," 2016) Even writing guides designed for literary studies students proclaim that "Paraphrasing and summarizing are usually superior to quoting…" (Gardner, 2004, p. 108). Similarly, Pecorari's (2006) assertion that "Quotation … comprises a significant minority of citations in the soft disciplines" (p. 10) would appear to overgeneralize the way sources are integrated into humanities and social sciences texts, consequently minimizing the central role direct quotation may play in specific disciplines. Such across-the-board assumptions about rhetorical strategies most certainly need rethinking given the studies are needed if we wish to more fully understand academic writing in the humanities.
Discipline-specific research is invaluable for both teachers and students of academic writing.
When advising students on textual norms and best practices, educators often must rely on generalizations that may not be applicable across a range of disciplines; in order to provide effective instruction, EAP practitioners working with both L1 and L2 learners need to present solid evidence of how successful writers communicate with their target discourse communities. In addition, increasing our awareness and knowledge of how writing varies across the disciplines will no doubt productively inform our pedagogical practices. Graduate writers, for their part, would benefit from exposure to research into writing in their field so that they may emulate norms; with regard to incorporating outside sources into their work, this means being able to look to the precedents of their peers in questions ranging from verb choice to whether to favour integral or non-integral forms.
Beyond providing exemplary patterns of writing, engaging with such investigations can bring to the fore a multiplicity of issues surrounding working with sources and the rhetorical impact of their writing choices. On a stylistic level, recognizing, for example, that multiple syntactic constructions for integral citations are being used in scholarly work could potentially add variety to the novice writer's palette. My own classroom experience confirms that when graduate students are exposed to research into writing in their field, they start thinking about writing in a much more in-depth manner.
One of our objectives as EAP educators is to help graduate students attain their publication goals and thereby disseminate important new research; it is hoped that this paper and future corpus-based research into discipline-specific writing practices will provide us with tools we can use to achieve those ends.