Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of English for Academic Purposes 37 (2019) 11e21

Contents lists available at ScienceDirect

Journal of English for Academic Purposes


journal homepage: www.elsevier.com/locate/jeap

Authorial voice constructed in citation in literature reviews of


doctoral theses: Variations across training contexts
Jian-E Peng
College of Liberal Arts, Shantou University, 243 Da xue Road, Shantou, Guangdong, 515063, China

a r t i c l e i n f o a b s t r a c t

Article history: Constructing authorial voice is considered essential in English academic texts. Citation as an
Received 4 November 2017 important discursive feature contributing to authorial voice has been underexplored, despite
Received in revised form 17 April 2018 fruitful research on citation practices. It also remains largely unknown whether variations in
Accepted 1 November 2018
citation-based voice construction exist in doctoral theses completed across training contexts.
Available online 2 November 2018
This study explores and compares authorial voice constructed in citation in the literature
review chapters of 20 doctoral theses written by home-grown and overseas-trained Chinese
Keywords:
writers (HGWT and OTWT), complemented by interviews with four of the 20 writers. It was
Authorial voice
Citation
found that compared to the HGWT, the OTWT displayed significantly greater use of non-
Doctoral theses integral citations, “author as agent” integral forms, and summary that contribute to authorial
Training contexts voice. Descriptive statistics for the two subcorpora also suggested that authorial voice was
Discursive feature more often communicated via lexical devices loaded with evaluation by the ten home-grown
writers whereas the ten overseas-trained writers preferred syntactic and discoursal devices.
While the present results are more suggestive than deterministic, this study highlights the
importance of promoting novice writers' awareness of the obscure and often unnoticed role
of citation in constructing authorial voice.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

The construction of authorial voice in academic texts is considered essential (Hyland & Sancho-Guinda, 2012), yet despite its
ostensibly clear meaning, voice is an ambiguous term to define (Tardy, 2012). Generally, voice may be viewed as the repre-
sentation of one's point of view (Hyland & Sancho-Guinda, 2012) or the negotiation of one's discursive identity (Matsuda, 2001)
or academic visibility (John, 2012), mainly originating from Western rhetorical traditions (Ramanathan & Atkinson, 1999).
Constructing authorial voice in academic writing is often a challenging issue (John, 2012; Matsuda & Tardy, 2007), particularly
among writers using English as an additional language (EAL) (Flowerdew, 2001; Hirvela & Belcher, 2001).
Many discursive features can contribute to authorial voice, such as hedges, boosters, metadiscourse, and citation (Nelson &
Castello, 2012; Thompson, 2012). Citations allow writers to delineate their research space as important and distinct from the
literature within their disciplinary community. This requires the writer to handle their own voice and the voices of the cited
authors simultaneously.1 More accurately, citation involves double-voicing (Bakhtin, 1978), whereby the writer injects his or
her intention and position into the representation of others' work, so the cited work may contain both the writer's and the
author's voices. Given the “dialogism and intertextuality” embedded in citations (Swales, 2014, p. 119), different forms of

E-mail address: pengjiane@stu.edu.cn.


1
This study adopts Thompson and Ye’s (1991) practice of assigning “writer” and “author” respectively to the person who cites and the cited person.

https://doi.org/10.1016/j.jeap.2018.11.001
1475-1585/© 2018 Elsevier Ltd. All rights reserved.
12 J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21

citation, such as integral or non-integral citations and brief or long quotations, may arguably bear on the writer's voice. While
citation practices have been widely researched, their association with the construction of authorial voice has been largely
underexplored.
It is also unclear whether citation practices, and accordingly the construction of authorial voice in doctoral theses, varies
between EAL writers who are locally trained and those who received academic training overseas. Higher education has
witnessed a surge in international student mobility that favors English-speaking countries (Lynch, 2015). For instance, the
trend towards Chinese students studying abroad is increasing, with the UK, the USA, and Australia being the favorite des-
tinations (Ministry of Education, 2016). Recent research has reported gains in L2 writing resulting from studying in native
English-speaking universities (see Wu & Zhang, 2017). However, little is known about whether advanced L2 users (e.g.
doctoral candidates) studying in Western institutions will become acculturated into English academic conventions. Analysis
and comparison of voice construction in citation between doctoral theses written by overseas-trained and home-grown
writers may raise EAL students’ sensitivity to writing across academic contexts, and also promote their awareness of the
obscure and often unnoticed role of citation in communicating authorial voice in English academic texts.
This study aims to explore voice construction in doctoral theses written by native Chinese-speaking writers who received
doctoral training in mainland China (i.e. home-grown writers) or abroad (i.e. overseas-trained writers). Like their counter-
parts elsewhere, Chinese students have been found to experience difficulties with citation or source use (Davis, 2013; Shi,
2006), and by extension with voice construction. It is expected that this study may be informative for EAL writers in
similar contexts.

2. Authorial voice in academic writing

An identifiable voice is a typical characteristic of English academic texts (Groom, 2000), although opinions about the exact
meaning of voice can be as diverse as the voices themselves. Matsuda (2001) broadly defined voice as “the amalgamative
effect of the use of discursive and non-discursive features that language users choose, deliberately or otherwise, from socially
available yet ever-changing repertoires” (p. 40). This definition treats voice as an accumulative effect created by language
users, or more specifically by both the writer and the reader (Tardy & Matsuda, 2009), who intentionally or unintentionally
use clues when engaging in the text.
Voice is a multifaceted concept encompassing both individual and social dimensions (Tardy, 2012). A textual voice is
intuitively individual in that it represents one's unique features expressed in the written mode. The individual aspect of voice
represents an “individual imprint on a text” (Tardy, 2012, p. 35), associated with authenticity, authorial presence, or
commitment. The voice of a text is also social because the writing and interpretation of a text are always situated within a
particular social or disciplinary context that carries its values. Writers' choices of discursive features indicate their identifi-
cation with the surrounding social worlds; therefore, a writer's voice is “multiple and intertextual” (Ramanathan & Atkinson,
1999, p. 50). This means that all texts involve multiplicity of voice, or in Bakhtin’s (1981) term, heteroglossia, which comes
from “those individuals we read” or “the genres and discourses that we write within” (Tardy, 2012, p. 39). Therefore, a writer's
voice may develop from their authorial presence discursively established in the text, and their allegiance to the disciplinary
and social communities in which they are situated. Tardy (2012; see also Matsuda & Tardy, 2007) added a reader perspective,
proposing that voice is also co-constructed by readers since it is the reader's impressions that identify voice. Insightful as this
perspective is, without involving text reviewers, this study mainly addresses the individual and social aspects of voice.
While the importance of voice in academic writing is widely acknowledged, its evaluation may not be straightforward.
Whether particular discursive features and the resulting strength of voice decide the quality of writing is a question that
defies quick answers. It has been found that board members of academic journals constructed a writer's voice based on
neither discrete discursive nor non-discursive features (Tardy & Matsuda, 2009), and there was no link between “reviewers'
constructions of the author's voice and their respective evaluations of the manuscript” (Matsuda & Tardy, 2007, p. 247). John
(2012) also noted that “[i]t is not always the case that more visibility leads to a better text” (p. 198).
Among the discursive devices pertinent to voice, such as hedges, boosters, and metadiscourse (Tardy, 2012), citation has
received insufficient attention (see Groom, 2000; John, 2012; Thompson, 2012), notwithstanding the fruitful research on
citation practices. It is argued here that citations inextricably embody the writer's voice. As previously discussed, citation in-
volves the blending of multiple voices. The construction of the writer's voice is always in relation to other voices through averral
and attribution. Averral refers to representation of the writer as the person who takes responsibility for a proposition, either by
making no attributions or using first-person pronouns or comments to signal the writer's presence (Tadros, 1993). Attribution
refers to “reports in the text which have the effect of transferring responsibility for what is being said” (Tadros, 1993, p. 104).
Simply put, averral asserts the writer's own voice, and attribution foregrounds the voices of previous contributors from specific
disciplinary communities. Such intertextuality in citation renders the voice constructed therein both individual and social (see
Tardy, 2012). Citation can function as an important device for writers to project their voices while tacitly acknowledging and
evaluating contributions by others, so that their work is anchored in a network of preceding scholarship.
Recent research has explored voice as self-representation or the writer's identity, and has revealed voice construction in EAL
writers' first language (L1) discourse (e.g. Hirvela & Belcher, 2001; Hu & Cao, 2011; Matsuda, 2001). EAL writers are not voiceless
in their transition to English academic writing situations, but carry their L1 writing experiences as they wrestle with differences
between their L1 and English rhetoric (Hirvela & Belcher, 2001). More specifically, citation or source use may impose difficulties
on EAL writers who are trained in different rhetoric conventions (Davis, 2013). For instance, it has been documented that
J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21 13

Chinese students' writing practices in high school require little use of sources and their English writing at college level pri-
oritizes “the accuracy and complexity of language” (Hirvela & Du, 2013, p. 91). Thus they are often confused about why and how
to cite sources (Davis, 2013; Shi, 2006). EAL writers' inexperience in working with sources can complicate their awareness of
constructing authorial voice in citation. In their recent influential study drawing on systemic functional linguistics and
appraisal theory (see Coffin, 2009), Hu and Wang (2014) identified complex cross-discipline (applied linguistics vs. medicine)
and cross-linguistic (Chinese vs. English) variations in citation-based dialogic expansion (i.e. opening up the space for other
voices) and dialogic contraction (i.e. shutting down the space for other voices). In brief, it seems that compared with other
devices, citation may operate less explicitly in constructing voice, and is less readily acquired by novice writers.

3. Authorial voice constructed in citation

Built on previous research into citation practices (Hyland, 2002; Petric, 2012; Swales, 1990, 2014), this study concen-
trates on exploring how authorial voice is constructed in citation. The first aspect under consideration is integral versus
non-integral citations (Swales, 1990). Integral citations use the name of the cited author as a grammatical constituent of the
sentence, whereas in non-integral citations the cited author's name is either placed in parentheses or noted by a super-
script number (Swales, 1990). Accordingly, integral citations foreground the cited author, and non-integral citations
accentuate the reported proposition. In non-integral citations, the writer's voice is believed to be dominant and the author's
voice to be “a satellite component” of the writer's argument (Groom, 2000, p. 20), which creates “the effect of contracting
the dialogic space” (Hu & Wang, 2014, p. 23). Mansourizadeh and Ahmad (2011) also contended that non-integral citations
can convey a sense of objectivity and avoid blocking the flow of the discursive argument. In their study that compared the
citation practices of non-native English speaking experts and novice writers in Malaysia, they found that the novice writers
tended to use more integral citations in isolation, whereas the experts strategically used both integral and non-integral
citations without interrupting the flow of the text. Hu and Wang (2014) found greater use of integral citations in arti-
cles written in Chinese, and more non-integral citations in those in English. Besides such ethnolinguistic influences,
disciplinary variations were also identified, with hard disciplines favoring non-integral forms (Hyland, 1999; Thompson,
2001). In a recent study tracking diachronic changes over the past 50 years, Hyland and Jiang (2017) reported a growing
preference for non-integral forms in 360 research articles (RAs) published in journals from four disciplines (applied lin-
guistics, biology, engineering, and sociology). For instance, in applied linguistics non-integral forms rose from 29% of all
citations in 1965 to 73% in 2015.
The second aspect explored here is how cited works are incorporated into a text by means of direct quotations, sum-
mary of a single source, and generalization where “material is ascribed to two or more authors” (Hyland, 1999, p. 348).
Summary and generalization, which correspond to assimilation in Hu and Wang’s (2014) analytic framework, allow writers
to rephrase the cited propositions into their own textual context (Hyland & Jiang, 2017) and merge them with their voices
(Hu & Wang, 2014). This can open up a wider space for the writer's voice, but according to Hu and Wang (2014) it also
contracts the dialogic space for alternative views. In contrast, since quotations clearly express others' voices, their overuse,
especially of extended quotations which do not take an evaluating position, can marginalize the writer's own voice (Borg,
2000), which may result in “a tissue of quotes” with no argument from the writer (Baynham, 1999, p. 493). Notably, in Hu
and Wang’s (2014) framework, insertion, which is similar to quotation, is dialogically expansive in that it presents the cited
propositions as individual subjectivity that is open to negotiation. That said, careful use of quotations, such as using
quotation fragments instead of lengthy excerpts, may create space for the writer's voice or evaluation of the cited work. In
Petri
c’s (2012) study of eight high-rated and eight low-rated Master's theses in gender studies written in English, contrary
to common expectations, high-rated theses contained three times as many direct quotations as low-rated theses. It turned
out that the writers of the high-rated theses employed more quotation fragments that were blended into their discourse.
Petri
c (2012) suggested that this practice of selectively quoting stretches of text can raise the discourse “to a more expert
level” (p. 111). Pickard’s (1995) study also found that in 11 articles in applied linguistics, about one third of the citations
involved direct quotations, of which 50% contained only one or two words. Hu and Wang (2014) detected that insertion and
assimilation citations were more frequently used in the English RAs than in the Chinese ones. In applied linguistics,
quotations were found to account for 10% in Hyland’s (1999) study, a proportion which has grown continuously from 1965
to 2015 (Hyland & Jiang, 2017).
Reporting verbs used in citations often convey the writers' evaluation of the cited work, which can directly express
authorial voice. The evaluative functions of reporting verbs in citations have been widely recognized and analyzed (Hyland,
1999, 2002; Thomas & Hawes, 1994; Thompson & Ye, 1991). In their comprehensive discussion of authorial voice, Nelson and
 (2012) pointed out that writers’ attitudes toward the cited work can be ascribed in reporting verbs. Drawing on the
Castello
classification of reporting verbs proposed by Thompson and Ye (1991) and Thomas and Hawes (1994), Hyland (2002)
developed a succinct taxonomy for analyzing reporting verbs, which was adopted in this study and is presented in the
next section.
As argued above, voice construction is considered a thorny challenge for novice writers, and citation is a key discursive
feature contributing to voice. In a doctoral thesis the bulk of the citations usually appear in the literature review (LR)2 chapter,

2
In this study the literature review of a thesis is defined as the entire chapter or major section(s) between the Introduction and Methodology chapters.
14 J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21

in which the writer needs to synthesize relevant scholarly work and create a research niche (Swales, 1990). The present study
sets out to analyze variations in voice construction via citations between the LRs of doctoral theses written by home-grown
and overseas-trained writers, supplemented by interview data. This study was guided by the following research questions:

1) Are there differences in how authorial voice is constructed in integral and non-integral citations between the LRs in
doctoral theses written by home-grown and overseas-trained writers?
2) Are there differences in how authorial voice is constructed in citations adopting quotations, summary, and generalization
between the aforementioned subcorpora?
3) Are there differences in how authorial voice is constructed in citations using reporting verbs between the aforementioned
subcorpora?

4. Methods

4.1. The corpus and informants

The corpus consisted of the LRs of 20 doctoral theses written by native Chinese-speaking writers, ten of whom obtained
their doctoral degrees from universities in mainland China and ten from universities in the UK (n ¼ 4), the USA (n ¼ 3), or
Australia (n ¼ 3), which are the most popular host countries for outbound Chinese students (Ministry of Education, 2016). The
theses by overseas-trained writers were identified by the following three criteria: a) The writers’ names conformed to the
Pinyin spelling system of Mandarin Chinese; b) In the Acknowledgements the writers acknowledged their relatives (e.g.
partners or parents) and colleagues in mainland China (see Geng & Wharton, 2016); and c) The research was carried out with
participants in mainland China.3 The theses were completed in between 2014 and 2016 and were chosen from the domain of
linguistics and applied linguistics, because in China doctoral theses in other domains are mostly written in Chinese.
Of the 20 LRs, 17 were retrieved from the following online databases: China National Knowledge Infrastructure (CNKI), the
official academic database in China; the ProQuest Dissertations and Theses Database; British Library EThOS; and Australasian
Digital Thesis. In addition, since this study planned to invite thesis writers for interview and the archived theses contained
little contact information for their writers, the remaining three LRs were submitted by three newly graduated doctoral thesis
writers (home-grown, n ¼ 1; overseas-trained, n ¼ 2) who, together with another local writer whose work was archived in
CNKI, were invited to subsequent interviews. The 20 collected LRs comprised a corpus of 292,815 words with tables, graphics,
and footnotes excluded, of which the home-grown writers' theses (HGWT) amounted to 162,220 words and the overseas-
trained writers’ theses (OTWT) 130,595 words.
Of the four interviewees,4 Lin (male, aged 38) and Chen (female, aged 36) obtained their doctoral degrees in mainland China,
from a national key university and a provincial key university that specializes in foreign languages respectively. Lin and Chen
submitted their doctoral theses in 2014 and 2016. The other two, Guan (male, aged 34) and Yuan (female, aged 33), held doctoral
degrees from prestigious universities in the USA and the UK respectively, and both submitted their doctoral theses in 2016.

4.2. Analytic framework for the corpus

The analytic framework was built on a synthesis of previous citation research on integral and non-integral citations
(Swales, 1990, 2014), the presentation of cited work (Groom, 2000; Hyland, 1999, 2002), and reporting verbs (Hyland, 2002).
Extending Swales' (1990) distinction between integral and non-integral citations, this study followed Swales (2014) and
further divided integral citations into four types: a) Author as subject of the sentence; b) Author as agent (e.g. “It was proposed
by Smith (2002)“); c) Author as adjunct (e.g. “According to Smith (2002)” or “As Smith (2002) points out”); and d) Author in noun
phrase (NP) (e.g. “Smith's (2002) theory” or “The theory proposed by Smith (2002)“). Swales (2014) analyzed the Michigan
Corpus of Upper-level Student Papers and found that “author as subject” was the most common category, followed by “author
in NP”. Referencing Lancaster’s (2012) findings, Swales (2014) defined the “author in NP” type as concept-focused citation, in
contrast to the “author as subject” type which is person-focused, and he argued that concept-focused citations “tend to show
greater conceptual integration of the cited sources” (p. 133).
With reference to Hyland (1999), the presentation of cited work was analyzed from three aspects: direct quotation,
summary, and generalization. Direct quotation was further divided into three types, following Borg (2000): a) Fragments (i.e.
duplicated texts shorter than a T-unit5); b) Brief quotations (i.e. T-units less than 40 words); and c) Extended quotations (i.e.
quotations longer than 40 words that are indented as blocks). This taxonomy presents a detailed framework for analyzing
quotations in terms of their syntactic placement. It may be expected that quoting a fragment will retain more space for the

3
The chosen theses were cross-checked with combined information about the writers' English learning or teaching experience, PhD scholarship awarded
by China Scholarship Council, and personal bio, which were retrieved from the theses or online (e.g. websites of the writers' current institutions in China
and academic social media such as the Linkedin). The obtained information confirmed that these writers were native Chinese-speakers.
4
Pseudonyms are used throughout to preserve anonymity.
5
A T-unit is an independent clause with any subordinate structures attached to it (Borg, 2000).
J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21 15

Reporting

Research Acts Cognition Acts Discourse Acts

Findings Procedures Positive Critical Tentative Neutral Assurance Doubt Counters

factive counter-factive non-factive factive non-factive tentative critical

Fig. 1. Taxonomy of reporting verbs (reproduced from Hyland, 2002, p. 119, p. 119).

writer's voice than when a syntactically and semantically complete clause is quoted. According to this taxonomy, a single
sentence may contain two or more instances of quotation, which were coded.
Reporting verbs were analyzed based on Hyland’s (2002) taxonomy, which classifies reporting verbs into three categories
of acts according to the activities they refer to: a) Research acts (including Findings and Procedures), related to research
activities occurring in the real world; b) Cognition acts, associated with the researcher's mental process; and c) Discourse acts,
which “involve linguistic activities and focus on the verbal expression of cognitive or research activities” (Hyland, 2002, p.
118). Within each category, reporting verbs are delineated according to the evaluative potentials that writers exploit to ex-
press their stance towards the reported propositions. The Findings category of Research Acts includes factive verbs (e.g.
demonstrate) that show writers' acceptance of the cited propositions, counter-factive verbs (e.g. fail) with which writers
present the cited propositions as false or incorrect, and non-factive verbs (e.g. find) that convey no clear judgment on their
reliability. The Procedures category of Research Acts describes procedural aspects and does not bear attitudinal loadings.
Cognitive Acts include verbs that express four types of evaluation: a) Positive (e.g. agree), where writers portray the author
as showing positive attitude toward the cited matter; b) Critical (e.g. disagree), where writers portray the author as taking a
critical stance; c) Tentative (e.g. believe), where writers represent the author as holding a tentative view towards the reported
matter; and d) Neutral (e.g. conceive), where the author is characterized as staying neutral toward the proposition.
Discourse Acts contain verbs to express three types of attitudes: assurance (including factive and non-factive), doubt
(including tentative and critical), and counters. Within the “assurance” type, factive categories (e.g. point out) are used by
writers to represent the reported information as true, and non-factive categories simply inform readers without conveying
the writer's interpretation. Of the “doubt” type, tentative verbs express writers' tentative attitudes toward the reported in-
formation (e.g. suggest), and critical verbs convey writers' disapproving attitudes (e.g. not account for). The last type of
Discourse Acts is “counters”, which portray the author as the agent who objects to or challenges the truth or correctness of the
cited proposition. Put differently, in using “counters” the writer places the author as taking responsibility for these reser-
vations (Hyland, 2002). Hyland's (2002) taxonomy of reporting verbs is shown in Fig. 1.

4.3. Procedures

The 20 LRs were input to NVivo 8.0 for repeated reading, coding, and in-depth analysis. Instances of citation were
identified following these guidelines: a) Canonical citation forms such as a proper name followed with a date in parentheses;
and b) Citation forms where the Subject pronoun clearly refers to the same previously quoted author and the possessive forms
of the proper name was previously quoted (Thompson & Ye, 1991). However, when “groups or ‘schools’ of researchers and
scholars” were followed with no actual date in their first occurrence, as in “Chomsky and his co-workers have recently”, this
was not counted as a citation (Swales, 1990, p. 150).
Based on this analytic framework, citations in the LRs were coded in terms of citation forms, presentation of cited work,
and reporting verbs. However, categorization decisions for the reporting verbs were not easy, and may be inaccurate because
the categories are not watertight and overlap can occur between them (Thompson & Ye, 1991). To enhance analytic validity, a
native English-speaking professor who taught academic writing was later invited to check the classification of the verbs
generated from the analysis. He was presented with Hyland's (2002) paper, the definitions of the categories, and a list of 491
coded reporting verbs, and independently reviewed the classification. Initially he agreed with the classification of all the
reporting verbs except two. After we checked the original context, revision was made to the classification of one verb.
Descriptive statistics were calculated by counting raw frequencies of the categories. Because the two subcorpora were of
different sizes, raw frequencies were standardized per 1000 words to minimize any effect caused by differing corpora sizes
(see Petri
c, 2012). The Mann-Whitney test was then used to test differences in citation forms, presentation of cited work, and
reporting verbs between the subcorpora. Given the small sample size, the Fisher's exact test was run and the exact signifi-
cance was considered when determining the significance of any difference (Field, 2009). The effect size was calculated using
pffiffiffi
the Z score obtained from the Mann-Whitney test based on the formulation: r ¼ Z/ n (Field, 2009, p. 550). The absolute
values of the effect size represent the magnitudes of difference, which can be interpreted as: r ¼ 0.1 (small effect), r ¼ 0.3
(medium effect), r ¼ 0.5 (large effect) (Field, 2009, p. 57).
Based on the quantitative results, four semi-structured interviews were conducted in Mandarin Chinese with the four
invited writers, to elicit their thesis-writing experiences and perceptions of citation practices. Each interview lasted about
30 min and was recorded, transcribed and translated into English by the researcher. The transcripts were then content-
16 J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21

Table 1
Integral and non-integral citations.

Sub-corpus Non-integral Integral


HGWT % 48 52
Total raw counts (total converted counts) 1245 (7.67) 1339 (8.25)
OTWT % 53 47
Total raw counts (total converted counts) 1314 (10.06) 1185 (9.07)
Exact. sig. U ¼ 24*, p ¼ .05 U ¼ 44, p ¼ .68
Effect size r 0.44 0.10
*
p < .05.

analyzed using NVivo 8.0, with themes inductively emerging and being summarized from the textual data (see Do €rnyei,
2007). Due to space constraints, the interview data are presented here not as a stand-alone entity but to complement the
quantitative analysis.

5. Findings and discussion

5.1. Authorial voice constructed in integral and non-integral citations

The distributions of integral and non-integral citations across the subcorpora are shown in Table 1. It can be seen that the
standardized counts of both integral and non-integral citations in the OTWT (9.07 and 10.06 respectively) were larger than
those in the HGWT (8.25 and 7.67 respectively). However, only the difference in non-integral citations was significant (U ¼ 24,
p < .05, Fisher's exact test, r ¼ 0.44, medium effect). The larger proportion of non-integral citations in the OTWT mirrors the
greater instances of non-integral forms found in Mansourizadeh and Ahmad (2011) and Swales (2014), and is also in tune with
the increasing preference for non-integral forms reported in Hyland and Jiang (2017). In non-integral citations priority is
given to propositions rather than to the cited authors (Swales, 1990) and the writer's voice tends to be dominant (Groom,
2000). Also, the use of non-integral forms can avoid blocking the flow of a text (Mansourizadeh & Ahmad, 2011) and
thereby enhance coherence, and textual coherence, according to Thompson (2012), enables the writer to “develop a voice of
authority” (p. 125). Taken together, authorial voice appeared to be more strongly communicated in non-integral citations in
the theses written by overseas-trained writers than in those by home-grown writers.
Table 2 shows the distribution of the integral citation types. The standardized counts of “author as subject” in HGWT was
greater than that in OTWT, whereas the standardized counts of “author as agent”, “author in adjunct”, and “author in noun
phrase” were greater in OTWT than those in HGWT. However, only the difference in “author as agent” reached statistical
significance (U ¼ 20, p < .05, Fisher's exact test, r ¼ 0.46, medium effect).
Meanwhile, to risk oversimplification, the overuse of “author as subject” may weaken the writer's voice and result in a
“shopping list” of literature. In this sense, the lower incidence of “author as subject” citations in the OTWT (4.47) than in the
HGWT (5.30) might communicate stronger authorial voice. In the following Example (1)6, the three cited authors are the
syntactic subjects, and due to the absent of evaluative expressions, the three citations constitute mostly a list describing “who
did what”. According to Groom (2000), the use of neutral verbs “analyzed”, “explored”, and “presents” has “the effect (intended
or otherwise) of transferring the responsibility for the attributed proposition” (p. 18) to the authors, which imposes a textual
expectation of the writer's subsequent averral indicating agreement or disagreement. Student writers who fail to do this are
often accused of being too descriptive in their writing (Groom, 2000). In contrast, in (2), as indicated by the conjunctive adverbs
“furthermore” and “sometimes”, the writer presents his/her averral about the effect of teacher behavior, which were sum-
marized from the citations in the parentheses. The non-integral forms attach emphasis to the propositions which reflect the
writer's train of thought organized from the literature, and accordingly the writer's voice is more explicitly constructed.

(1) Inspired by White, O'Connella et al. (2005) have analyzed …. Juana et al. (2006) have explored and compared ….
Stenvall (2008) presents two complementary approaches … (HGWT 9)
(2) Furthermore, certain types of teacher behavior will cause students to feel isolated, such as … (Tutula, 2002). Some-
times, confrontations occur between teaching and learning styles as the conflict of the different … (Lucas & Miraflores,
2011). (OTWT 8)

In the interviews, Lin and Chen, who were the two home-grown writers, both mentioned that they mostly use “author as
subject” citations for purposes of emphasis or comparison. As Chen stated, “When there happened to be several authors who
raised different opinions, I would deliberately juxtapose their opinions with the authors’ names placed to the front.”
Potentially such knowledge-telling citations, if without ensuing evaluation from the writer, could be deemed descriptive. In

6
Highlights in bold and/or italics in all extracts are added for referencing purposes.
J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21 17

Table 2
Integral citation types.

Sub-corpus Subject Agent Adjunct Noun Phrase


HGWT % 64 4 11 21
Total raw counts (total converted counts) 859 (5.30) 57 (0.35) 141 (0.87) 282 (1.74)
OTWT % 49 7 14 30
Total raw counts (total converted counts) 584 (4.47) 83 (0.64) 160 (2.16) 358 (2.74)
Exact. sig. U ¼ 45, p ¼ .74 U ¼ 20*, p ¼ .02 U ¼ 31, p ¼ .17 U ¼ 31, p ¼ .17
Effect size r 0.08 0.51 0.32 -.32
*
p < .05.

comparison, the account of Yuan, who was an overseas-trained writer, enunciated her subtle treatment of double-voicing in
citation when handling integral and non-integral forms:
The subject should not keep shifting … For instance, if in this paragraph I am talking about the nature of critical
thinking, I will use critical thinking or its pronoun as the subject of many sentences, instead of writing ‘somebody said
something’.

5.2. Authorial voice constructed in the presentation of cited work

The HGWT and OTWT subcorpora were found to display differences in the way cited work was presented, although many
of these differences did not reach significant levels. Table 3 shows the results.
As can be seen, the majority of cited work was represented as a summary in the two subcorpora, which is consistent with
Hyland's (1999) findings, although Hyland and Jiang (2017) reported a rise in generalization. The converted count for sum-
mary in the OTWT (13.11) was significantly larger than in the HGWT (11.00) (U ¼ 23, p < .05, Fisher's exact test, r ¼ 0.46,
medium effect). Summary and generalization are dialogically contractive (Hu & Wang, 2014) with the writer gaining greater
flexibility to tailor citations to support his or her claims, thereby maybe augmenting authorial voice. English RAs were found
to contain more assimilation (i.e. summary and generalization) than their Chinese counterparts (Hu & Wang, 2014). In this
light, the greater frequency of summary identified in the OTWT may suggest that the overseas-trained writers were more
acculturated into this English discursive feature. As illustrated in (3), summary (with relevant stretches of texts shown in bold
italics) and generalization (with relevant stretches of texts shown in bold) were used to weave several authors' works
together to support the writer's argument, and the text unfolds effectively so that the writer's voice is clearly expressed
among heteroglossia (Bakhtin, 1981) built on the network of the citations.

(3) Though Brown and Levinson's model has been used as a theoretical premise for different researchers (e.g. Ralarala,
2007), controversy in eastern and western approaches to politeness is prevalent (c.f. Lang, 1998; Ye, 2004). In
particular, Ye (2004) claims that … There are researchers who examine … (Gu, 1990; Hu, 1944; Mao, 1994). It has
been argued that Brown and Levinson's model … (Mao, 1994). (OTWT 1)

It should be noted that generalization occupies the smallest percentage (18%) in OTWT, which was dissimilar from Hyland
and Jiang’s (2017) reported increased use of generalization in citations. The reason may be attributed to the different genres
examined. The present study examined LRs in doctoral theses. This sub-genre often obliges the writer to engage in substantial
reviews of important work in the field. Hyland and Jiang’s (2017) report targeted RAs that have seemingly trended towards
the “nonreporting nonintegral citing style” akin to “parenthetical plonking” or “nods all round to previous researchers”
(Swales, 2014, p. 135). Consequently, a wide coverage of sources within the space allowed in journals entails frequent use of
generalization.
Table 4 presents the types of quotations, the number of words quoted in the three types of quotations, and mean length of
direct quotations in number of words. As seen in this table, the number of words quoted per 1000 words in the HGWT was
44.43, which was smaller than in the OTWT (52.47). However, the mean length of direct quotations (i.e. quoted words/
number of direct quotations) for the HGWT (16.69) was larger than that for the OTWT (12.28). This contrast was similar to
what Petric (2012) found in low-rated and high-rated Master's theses, which revealed a larger number of words quoted per
1000 words and a shorter mean length of direct quotations in the high-rated theses than in the low-rated theses.
Table 4 also shows that in both subcorpora quotation fragments accounted for the largest proportion, followed by a small
number of brief quotations and negligible instances of extended quotations. The converted counts of quotation fragments in
the HGWT (1.87) was smaller than in the OTWT (3.28), whereas more extended quotations were found in the HGWT (0.24)
than in the OTWT (0.20), although these differences were not statistically significant. Petri c (2012) also reported more
quotation fragments in the high-rated theses than in the low-rated theses. It may be suggested that the ten OTWT in this
study maintained a more skillful control of double-voicing in quotations by using short quotes more frequently and fewer
extended quotes. This quoting style, according to Petric (2012), can create “a mosaic out of the appropriated material and
one's own words”, which is legitimate and reflects “a higher level of academic literacy in comparison to patchwriting” (p. 111).
18 J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21

Table 3
Presentation of cited work.

Sub-corpus Direct Quotations Summary Generalization


HGWT % 16 66 18
Total raw counts (total converted counts) 437 (2.69) 1785 (11.00) 496 (3.06)
OTWT % 20 62 18
Total raw counts (total converted counts) 558 (4.27) 1712 (13.11) 495 (3.79)
Exact. sig. U ¼ 42, p ¼ .58 U ¼ 23*, p ¼ .04 U ¼ 36, p ¼ .28
Effect size r 0.14 0.46 0.25
*
p < .05.

Table 4
Types of direct quotations.

Sub- Quotation Brief Extended Number of words Mean length of


corpus fragment quotation quotation quoted DQ
HGWT % 69 22 9
Total raw counts (total converted 303 (1.87) 95 (0.59) 39 (0.24) 7207 (44.43) 16.49
counts)
OTWT % 77 18 5
Total raw counts (total converted 428 (3.28) 104 (0.80) 26 (0.20) 6852 (52.47) 12.28
counts)
Exact sig. U ¼ 39, p ¼ .44 U ¼ 48, p ¼ .91 U ¼ 43.5, p ¼ .64 U ¼ 49, p ¼ .97 U ¼ 34, p ¼ .25
Effect size 0.19 0.03 0.11 0.02 0.27
r

As seen in (4), quotation fragments taken from the same source, together with a summary (shown in bold italics), were
integrated into the writer's averred statement:

(4) According to Davies and Barnett, criticality comprises ‘thinking, being and acting’ (2015, p.15), meaning that it is not
enough to …, but one also needs to act critically (Moon, 2008), to participate in society and to act ‘ethically on the
basis of reasoned judgments’ (Davies and Barnett, 2015, p.16). (OTWT 7)

The interview data also showed that the four writers used direct quotations to retain the authenticity of and/or to
emphasize the quoted content. For instance, Lin articulated that he used quotations “when the content is important, and
when summary of it may cause misunderstanding”. However, Yuan's account also revealed that she would strategically blend
fragments from sources to her discourse context. As she reported, when handling citations, “I will pick the key elements in
somebody's accounts and place them as parts of a sentence, and put the name of the author in the parentheses.”

5.3. Authorial voice constructed in citations using reporting verbs

The Mann-Whitney tests showed no statistical differences in the use of reporting verbs between the subcorpora (exact
sig.>0.05); these results are not shown for reasons of space. Corresponding to Figs. 1 and 2 shows the distribution of the
reporting verbs.

Reporting

Research Acts Cognition Acts Discourse Acts


HGWT 44.81% 10.21% 44.98%
OTWT 46.83% 8.33% 44.84%

Findings Procedures Positive Critical Tentative Neutral Assurance Doubt Counters


HGWT 34.39% 65.61% 41.62% 2.31% 19.65% 36.42% 77.03% 19.95% 3.02%
OTWT 44.72% 55.28% 34.51% 1.77% 22.13% 41.59% 74.18% 24.01% 1.81%

factive counter-factive non-factive factive non-factive tentative critical


HGWT 37.93% 4.22% 57.85% 45.49% 54.51% 94.08% 5.92%
OTWT 26.76% 2.82% 70.42% 40.13% 59.87% 96.58% 3.42%

Fig. 2. Overall distribution of reporting verbs.


J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21 19

As seen in Fig. 2, research verbs and discourse verbs took up about 90% of the reporting verbs. The proportions of research verbs
in the two subcorpora (44.81% and 46.83%) were larger than in the RAs in applied linguistics (30.5%) reported by Hyland (1999),
whereas the discourse verbs in the subcorpora (44.98% and 44.84%) occurred less frequently (59%) than in Hyland (1999). This
may be because the LRs in doctoral theses involved a much lengthier review of previous research, and hence more research verbs
were used to recount research activities. However, these findings confirm the increasing preference for research verbs found in
Hyland and Jiang (2017), indicating that thesis writers tend to state their propositions with reference to real-world evidence.
Although no statistically significant difference was obtained, a consistent pattern seemed to emerge across the three
categories: The HGWT contained more verbs carrying the writers’ positive and negative evaluations of the cited works or
propositions. For instance, in the Findings category, factive and counter-factive verbs accounted for a larger percentage
(37.93% and 4.22% respectively) in the HGWT than in the OTWT (26.76% and 2.82% respectively). Examples of these two
subtypes are shown in (5) and (6). It is important to note that the phrase “fail to”, as in (6), occurred four times in the HGWT
but zero times in the OTWT.

(5) As Dewaele (2007, 2008) confirmed, CA/FLA levels are determined by … (OTWT 8)
(6) For example, whereas Chandler (2003) concludes that …, Kepner (1991), Polio et al. (1998), and Sheppard (1992) fail to
find … (HGWT 4)

Similarly, regarding cognition verbs, the HGWT contained a larger percentage of verbs portraying the cited authors as
having a positive attitude toward the proposition (the “positive” subtype) and taking a critical stance (the “critical” subtype)
(41.62% and 2.31% respectively) than the OTWT (34.51% and 1.77% respectively):

(7) Rene Descartes (2002) held that all the ideas required for … (HGWT 8)
(8) The Subset Principle is not believed to operate in SLA (see, e.g., Akiyama 2002). (HGWT 3)

Consistently, discourse verbs expressing the writers' endorsement of cited material (the “factive” subtype in “Assurance”)
and their critical evaluation (the “critical” subtype in “Doubt”) were used more frequently in the HGWT (45.49% and 5.92%
respectively) than in the OTWT (40.13% and 3.42% respectively). Verbs in “counters”, which place the evaluation responsibility
on the authors, also occurred more frequently in the HGWT (3.02%) than in the OTWT (1.81%). For instance, of the “counters”
verbs, criticize and reject were used most frequently in the HGWT (see Example 9), appearing nine times and three times
respectively, while in the OTWT the most frequently used counters are criticize and challenge (see Example 10), both occurring
four times. Overall, the comparison of the sub-corpora suggests that the OTWT's greater use of neutral cognitive verbs (see
Example 11) and the HGWT's higher incidence of positive cognition verbs and factive discourse verbs (see Example 12) bear
out the trend in RAs (Hyland & Jiang, 2017), whereas the HGWT's greater use of critical verbs somewhat contradicts the trend
of sparing use of critical verbs in RAs (Hyland & Jiang, 2017).

(9) Oakeshott (1962) also criticized the dominance of rationalism … (HGWT 8)


(10) In this sense, Engestrom (2006) challenges the traditional … (OTWT 10)
(11) Questionnaire was regarded by Dornyei (2003) as relatively easy to administer to … (OTWT 2)
(12) Schegloff (1990) points out that topics are emergent in talk and … (HGWT 2)

Although the non-significant differences forbid any generalization, the higher frequency of verbs showing positive and
critical evaluation in the HWGT suggests that the authorial voice embodied in reporting verbs was stronger in the ten HGWT
than in the ten OTWT, although stronger voice does not necessarily lead to better writing (John, 2012). The relatively higher
occurrence of critical verbs in the HGWT may reflect acceptable discursive practices in the home-grown writers' local aca-
demic community. As reported in Hu and Cao (2011), Chinese discourse contains more authorial certainty and assertiveness
than English discourse. This rhetoric convention could influence L1 Chinese writers’ lexical choices in English writing,
especially when they need to claim credit for their thesis in this “high-stakes genre” (Geng & Wharton, 2016, p. 80).
The above speculation was confirmed in the interviews. Of the two home-grown writers, Lin believed that verbs or phrases
such as fail to, misuse, and suffer from could and should be used in order to showcase the writer's knowledge of limitations or
mistakes in previous research. He articulated, “I think only when we identify those limitations or mistakes can our research be
justified. So theoretically, it is unusual not to use these expressions.” Chen expressed her understanding of necessary
politeness or euphemism when evaluating previous research, but she admitted using phrases like fail to in writing. In contrast,
of the two overseas-trained writers, Guan reported that when addressing research limitations, he tended not to use these
expressions or use the author as the subject, so as not to threaten the author's face. Yuan also avoided the use of such ex-
pressions because they impose the writer's opinions on the cited authors:
I don't easily use fail, because when we are evaluating other people's research, we are using our perspectives, and
people taking different perspectives would evaluate differently. Also, a study may accomplish one thing and it is
impossible that it could accomplish all the things you expected, so I didn't say it failed; instead, I would say it did this
but not that, to stay objective.
20 J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21

The interview data also indicated that the writers' citation practices were influenced by their interaction with the local
academic community and academic artifacts such as existing doctoral theses. For instance, Yuan recalled an incident where
she criticized previous research based on speculation and reasoning, and her supervisor suggested that she avoid such strong
claims before she obtained empirical evidence. Such interaction with her supervisor might have instilled in Yuan a mind-
fulness of expressing oneself in ways acceptable to the community. Besides acknowledging guidance from their supervisors,
the two home-grown writers both mentioned that they deliberately consulted existing doctoral theses not only from China
but also from English-speaking countries such as the USA and Australia. When asked why, Chen elaborated, “Since my major
was English and I was writing my thesis in English, the standard that I pursued was close to native speakers' standard, or to
meet the writing requirements of international universities.” Chen's narrative suggested the possible influence of English
academic artifacts on EAL writers' academic writing despite their physical distance from English-speaking communities. This
may partly explain the several non-significant differences between the subcorpora found in this study.

6. Overall discussion and conclusion

This study has explored how authorial voice was constructed in citation in the LRs of doctoral theses written by native
Chinese writers who received academic training either at home or abroad. Inferential statistics showed that the OTWT
displayed significantly greater use of non-integral citations, “author as agent” in integral forms, and summary than the
HGWT, which suggests that the OTWT communicated more prominent authorial voice by means of the writers' orches-
tration of cited materials and suppressing agency in citations. In addition, according to the descriptive statistics the OTWT
used more quotation fragments and fewer extended quotations, which also functions to construct a stronger voice since the
cited author's voice was more dispersed in the text and merged into the writer's averred statements. In contrast, more
positive and critical reporting verbs were found in the HGWT than in the OTWT, which indicates that the ten home-grown
writers expressed stronger authorial voice than their overseas-trained counterparts in the use of reporting verbs. These
findings may point to a pattern of difference between the two groups of writers: authorial voice was more often
communicated via lexical devices loaded with evaluation by the ten home-grown writers, whereas the ten overseas-
trained writers preferred syntactic and discoursal devices. This interpretation is restricted to the current corpus and re-
quires more evidence from future research. Overall, however, the findings for the OTWT, compared to the HGWT, were in
agreement with Hyland and Jiang (2017), who reported trends towards writers preferring non-integral forms, research
verbs, and non-evaluative structures.
The 20 LRs from doctoral theses written by writers of the same ethnolinguistic descent were found to exhibit certain
variations in citation-based voice construction across academic training contexts; some of these variations were significant
and others were only plausible to the current corpus. The present findings suggest that novice EAL researchers receiving
training in English-speaking contexts may be more likely to socialize into the “arcane conventions of academic discourse” in
English (Hyland, 2016, p. 62) during participation in their transplanted community of practice, as reported in the interviews.
Geng and Wharton (2016) also contended that specific genre proficiency is primarily developed “via the PhD journey as
undertaken within a specific institutional context” (p. 89).
On the other hand, some non-significant differences between the two subcorpora may be attributed to at least two reasons.
The first was the small sample size, which might not be able to reveal possible substantial differences in these respects. The
second may be to do with the thesis writers in this study. Doctoral candidates in applied linguistics are advanced English
language users, and even located in a non-English speaking context, they can develop academic skills through interacting with
local academic communities and consulting English-medium artifacts, as reported by the present informants.
Certain implications for teaching citation practices can be derived from the findings. EAP teachers should raise learners'
awareness of the subtleties of citation-based voice construction. Poor awareness in this regard is a perennial problem for L2
writers (John, 2012). Teachers could first require students to analyze the citation practices within their disciplinary com-
munity and the authorial voice embedded therein. Particular discussions can be directed on circumstances where authorial
voice needs to be explicitly projected or discursively implicated and the possible reasons, such as the writer's identity (e.g. an
expert or novice) or rhetorical expectations of a specific academic community. Then students could be supplied with texts
containing differing citation forms (e.g. many integral citations or direct quotations) and asked to study the variations in
authorial voice constructed in different citation forms. Instructors could engage students in process-writing tasks to make
guided revisions of their citations (see John, 2012). In particular, EAL students should be trained to realize that citing is not
simply a required behavior in academic writing, but an essential skill of “intertextual storytelling” (Swales, 2014, p. 133). This
means that they should learn to make informed decisions about how to cite, in order to guarantee their visibility in the text
while establishing associations between sources.
While this study has presented in-depth analyses, there is no attempt to establish direct causal relations between different
forms of citations and authorial voice. The identification of differing voices as constructed in citations is inevitably inter-
pretive and also to a large extent construed by the reader (Tardy, 2012). Groom (2000) likewise cautioned that “form-function
correlations” should not be seen “in any deterministic sense” (p. 22). Moreover, while variations were identified between the
subcorpora, many factors other than the training contexts, such as individuals’ prior academic experience or English profi-
ciency, could simultaneously contribute to these variations. However, despite its limitations this study has contributed a
nuanced empirical evidence of citation-based voice construction, which may inform EAL writers about the implications of
J.-E. Peng / Journal of English for Academic Purposes 37 (2019) 11e21 21

constructing authorial voice in various forms of citation, thereby negotiating their identity in specific disciplinary
communities.

Acknowledgements

I would like to thank the two anonymous reviewers for their constructive and helpful comments on the earlier version of
this paper. This study was supported by the National Social Science Fund of China (Project No. 14BYY067).

References

Bakhtin, M. M. (1978). Discourse typology in prose. In L. Matejka, & K. Pomorska (Eds.), Readings in Russian poetics: Formalist and structuralist views (pp.
176e196). Ann Arbor, MI: Michigan Slavic Publications.
Bakhtin, M. M. (1981). In M. M. Bakhtin, M. Holquist, C. Emerson, & M. Holquist (Eds.), The dialogic imagination: Four essays by. Austin: University of Texas
Press.
Baynham, M. (1999). Double-voicing and the scholarly ‘I’: On incorporating the words of others in academic discourse. Text, 19(4), 485e504.
Borg, E. (2000). Citation practices in academic writing. In P. Thompson (Ed.), Patterns and perspectives: Insights into EAP writing practice (pp. 26e43). Reading,
UK: Centre for Applied Language Studies.
Coffin, C. (2009). Incorporating and evaluating voices in a film studies thesis. Writing & Pedagogy, 1, 163e193.
Davis, M. (2013). The development of source use by international postgraduate students. Journal of English for Academic Purposes, 12(2), 125e135.
Do€ rnyei, Z. (2007). Research methods in applied linguistics: Quantitative, qualitative, and mixed methodologies. Oxford, UK: Oxford University Press.
Field, A. (2009). Discovering statistics using SPSS (3rd ed.). London: Sage.
Flowerdew, J. (2001). Attitudes of journal editors to non-native speaker contribution. Tesol Quarterly, 35(1), 121e150.
Geng, Y., & Wharton, S. (2016). Evaluative language in discussion sections of doctoral theses: Similarities and differences between L1 Chinese and L1 English
writers. Journal of English for Academic Purposes, 22, 80e91.
Groom, N. (2000). Attribution and averral revisited: Three perspectives on manifest intertextuality in academic writing. In P. Thompson (Ed.), Patterns and
perspectives: Insights into EAP writing practice (pp. 14e25). Reading, UK: Centre for Applied Language Studies.
Hirvela, A., & Belcher, D. (2001). Coming back to voice: The multiple voices and identities of mature multilingual writers. Journal of Second Language Writing,
10(1e2), 83e106.
Hirvela, A., & Du, Q. (2013). “Why am I paraphrasing?”: Undergraduate ESL writers' engagement with source-based academic writing and reading. Journal of
English for Academic Purposes, 12(2), 87e98.
Hu, G., & Cao, F. (2011). Hedging and boosting in abstracts of applied linguistics articles: A comparative study of English- and Chinese-medium journals.
Journal of Pragmatics, 43, 2795e2809.
Hu, G., & Wang, G. (2014). Disciplinary and ethnolinguistic influences on citation in research articles. Journal of English for Academic Purposes, 14, 14e28.
Hyland, K. (1999). Academic attribution: Citation and the construction of disciplinary knowledge. Applied Linguistics, 20(3), 341e367.
Hyland, K. (2002). Activity and evaluation: Reporting practices in academic writing. In J. Flowerdew (Ed.), Academic discourse (pp. 115e130). London, UK:
Longman.
Hyland, K., & Jiang, F. (2017). Points of reference: Changing patterns of academic citation. Applied Linguistics. Epub ahead of print, 31 May 2017.
Hyland, K., & Sancho-Guinda, C. (Eds.). (2012). Stance and voice in written academic genre. Basingstoke: Palgrave MacMillan.
John, S. (2012). Identity without ‘I’: A study of citation sequences and writer identity in literature review sections of dissertations. In R. Tang (Ed.), Academic
writing in a second or foreign language: Issues and challenges facing ESL/EFL academic writers in higher education contexts (pp. 186e203). London:
Continuum.
Lancaster, Z. (2012). Stance and reader positioning in upper-level student writing in political theory and economics. Unpublished doctoral dissertation. Ann
Arbor: University of Michigan.
Lynch, K. (2015). Control by numbers: New managerialism and ranking in higher education. Critical Studies in Education, 56(2), 190e207.
Mansourizadeh, K., & Ahmad, U. K. (2011). Citation practices among non-native expert and novice scientific writers. Journal of English for Academic Purposes,
10(3), 152e161.
Matsuda, P. K. (2001). Voice in Japanese written discourse: Implications for second language writing. Journal of Second Language Writing, 10(1e2), 35e53.
Matsuda, P. K., & Tardy, C. M. (2007). Voice in academic writing: The rhetorical construction of author identity in blind manuscript review. English for Specific
Purposes, 26(2), 235e249.
Ministry of Education. (2016). Briefing of blue book of the employment situation of Chinese overseas returnees 2016 [zhongguo liuxue huiguo jiuye
lanpishu 2016 qingkuang jieshao]. http://www.moe.edu.cn/jyb_xwfb/xw_fbh/moe_2069/xwfbh_2016n/xwfb_160325_01/160325_sfcl01/201603/
t20160325_235214.html.
Nelson, N., & Castello , M. (2012). Academic writing and authorial voice. In M. Castello , & C. Donahue (Eds.), University writing: Selves and texts in academic
societies (pp. 33e52). Bingley, UK: Emerald.
Petric, B. (2012). Legitimate textual borrowing: Direct quotation in L2 student writing. Journal of Second Language Writing, 21(2), 102e117.
Pickard, V. (1995). Citing previous writers: What can we say instead of ‘say’? Hong Kong Papers in Linguistics and Language Teaching, 18, 89e102.
Ramanathan, V., & Atkinson, D. (1999). Individualism, academic writing, and ESL writers. Journal of Second Language Writing, 8(1), 45e75.
Shi, L. (2006). Cultural backgrounds and textual appropriation. Language Awareness, 15(4), 264e282.
Swales, J. M. (1990). Genre analysis: English in academic and research settings. Cambridge, UK: Cambridge University Press.
Swales, J. M. (2014). Variation in citational practice in a corpus of student biology papers: From parenthetical plonking to intertextual storytelling. Written
Communication, 31(1), 118e141.
Tadros, A. (1993). The pragmatics of text averral and attribution in academic texts. In M. Hoey (Ed.), Data, description, discourse (pp. 98e114). London:
HarperCollins.
Tardy, C. M. (2012). Current conceptions of voice. In K. Hyland, & C. Sancho-Guinda (Eds.), Stance and voice in written academic genre (pp. 34e48).
Basingstoke: Palgrave MacMillan.
Tardy, C. M., & Matsuda, P. K. (2009). The construction of author voice by editorial board members. Written Communication, 26(1), 32e52.
Thomas, S., & Hawes, T. P. (1994). Reporting verbs in medical journal articles. English for Specific Purposes, 13(2), 129e148.
Thompson, P. (2001). A pedagogically-motivated corpus-based examination of PhD theses: Macrostructure, citation practices and uses of modal verbs. Un-
published doctoral thesis. Reading, Berkshire, UK: University of Reading.
Thompson, P. (2012). Achieving a voice of authority in PhD theses. In K. Hyland, & C. Sancho-Guinda (Eds.), Stance and voice in written academic genre (pp.
119e133). Basingstoke: Palgrave MacMillan.
Thompson, G., & Ye, Y. (1991). Evaluation in the reporting verbs used in academic papers. Applied Linguistics, 12(4), 365e382.
Wu, H., & Zhang, L. J. (2017). Effects of different language environments on Chinese graduate students' perceptions of English writing and their writing
performance. System, 65, 164e173.

Jian-E Peng is a professor in the College of Liberal Arts, Shantou University. She holds a PhD from the University of Sydney. Her research interests include
learner motivation, English academic writing, discourse analysis, teacher development, and research methodology.

You might also like