Previous studies on metadiscourse in academic writing (Alonso-Almeida, 2014; Belyakova, 2017; Cmejrkova, 1996; Hryniuk, 2018; Hu & Cao, 2011; Isık-Tas, 2018 Lee & Casal, 2014; Lee & Deakin, 2016; Mikolaychik, 2019; Pisanski Peterlin, 2005; Pyankova, 1994; Vassileva, 2001; Walková, 2018) have offered some interesting insights into cultural differences in academic writing. However, taking prior research altogether, it seems that English academic texts produced by L2 writers have been mainly analysed only in terms of their distinction from L1 academic discourse, and little attention has been paid to differences in metadiscourse features used by L2 writers exposed to different writing styles.

Pyankova (1994), for example, who explored differences between English and Russian academic texts, found that Russian scholars underuse self-mentions and overuse passive and impersonal structures. The article by Duszak (1994, p. 291), that concerned with the attempts to reveal differences in metadiscourse patterns used in academic prose by L1 and L2 (Polish) writers, explained these differences by «a history of socialization of academics to different discourse communities». In the same vein, Cmejrkova (1996, p. 137) revealed that Czech linguists writing in English were more «reluctant to commit themselves early to an announcement of the research purpose and preferred indirect declarations or rhetorical questions». Another study conducted by Vassileva (2001) on commitment and detachment patterns in English and Bulgarian linguistics research articles revealed differences in hedging and boosting tools that were explained by different rhetorical traditions.

Pisanski Peterlin (2005) conducted a contrastive analysis focusing on variation in the use of metadiscourse in English and Slovene research articles and found that metadiscourse devices were more restricted in Slovene academic writing than in English academic prose. Belyakova (2017) carried out a cross-cultural comparison of RA abstracts by L2 (Slavic) and L1 academic writers in geoscience to investigate their metadiscourse choices and found that writers from East European academia usually disguise themselves to a larger extent. One more study of metadiscourse in research articles was conducted by Hryniuk (2018) who explored how British and Polish writers represent themselves in academic discourse and investigated differences in frequencies and functions of the first person pronouns in applied linguistics RAs. The results showed that Polish scholars employed less first person pronouns and did not assume responsibility for what is stated. According to Walková (2018), who explored how L1 and L2 (Slovak) writers position themselves in research papers, Anglophone academic culture is rather individualistic with a predominance of the reader-inclusive perspective and the use of the first-person singular by single authors. The same results were obtained by Bogdanović and Mirović (2018) who compared Serbian and English-medium RAs written by Serbian authors.

In the Asian context, the same conclusions were drawn by Hu and Cao (2011), who explored the use of hedging and boosting devices in RA abstracts collected from Chinese- and English-medium applied linguistics journals. They found that English RA abstracts contained more hedges than Chinese ones. Li and Xu (2020), who analysed metadiscourse in sociology articles by and L1 and L2 (Chinese) writers, revealed that English writers used metadiscourse markers far more than their Chinese counterparts. Kustyasari et al. (2021) compared metadiscourse functions in research articles by native English and Malay writers. What is interesting, they revealed that in Indonesian-authored academic prose interactional and interactive metadiscourse markers performed similar functions, i.e. they were used to indicate a relation between sentences, involve readers in a dialogue, limit commitment to propositions, emphasize certainty, signal their attitudes towards claims and readers. Azar et al. (2022) made an attempt to compare hedges and boosters in L1 and L2 (Malay) research articles focusing on the «Introduction» section. They established notable differences in the use of these features which prevailed in L1 writers’ discourse. All these studies considered cultural backgrounds as the main factor influencing the choice of metadiscourse devices and ignored the role of English language proficiency.

In the context of globalization of science and education, new academic genres that penetrate into local educational contexts play a crucial role. One of these genres is research proposal (RP) which is a concise comprehensive description of future research including justification of the chosen research project, description of the current state of knowledge on the research issue, analysis of relevant literature, formulation of the hypothesis, definition of the research methodology, and formulation of expected results. RPs are written by students when applying to degree programs or for research funding, or they can serve as an educational task intended to assess the student’s ability to construct a research project in a discipline additionally providing an idea of student’s research interests and intent. RP writing is therefore an important task which can help in assessing the degree of development of both the linguistic and research competences in postgraduate students.

Despite the crucial role of this genre in English for Academic Purposes teaching, studies conducted on RPs are scarce, and mainly focus on grant proposals (Connor, 2000; Connor & Mauranen, 1999; Fazel & Shi, 2015; Feng, 2006; Feng & Shi, 2004; Flowerdew, 2016; Matzler, 2021; Pascual & Unger, 2010; Tardy, 2003; Tseng, 2011).

One of the first grant proposal studies was Connor and Mauranen’s research conducted on move-structures (Connor & Mauranen, 1999). For the research purpose, Connor and Mauranen analyzed English grant proposals written by Finnish researchers and identified ten rhetorical moves. In Connor’s (2000) further study on research grant proposals, she explored the transferability of this move-structure framework to grant proposals written by US researchers and found that different requirements of the US grant system caused structural variations in her corpus. Also focusing on the move structure, Feng and Shi (2004) analyzed English grant proposals submitted by Chinese researchers to a Canadian research agency and found that they lack the Compliance Claim move identified by Connor and Mauranen, but include the Communication of Results move that had not been identified in previous research. Another Feng’s (2006) study offered a generic description of Hong Kong Competitive Earmarked grant proposals by integrating lexical and pragmatic analyses with rhetorical move analysis. The study identified the most frequently occurring hedges and boosters, among which the modals may and will. Additionally, she found the frequent use of other hedges such as approximators - about, approximately and around; probability markers - likely; and words expressing usuality - typically, usually. In another study, Pascual and Unger (2010) investigated how Argentinian research grant proposal writers in chemistry and physics position themselves in relation to other researchers. The results indicated that the corpus featured a variety of metadiscourse devices used to invite rather than challenge colleagues’ views. Fazel and Shi’s (2015) study explored grant proposals from a different perspective making an attempt to reveal how L2 doctoral students use citations and what citation strategies help project their emerging scholarly identities.

From a cross-disciplinary approach, Nuzha et al.’s (2020) study intended to contrast data on the frequency of rhetorical moves used in proposals from two disciplines revealed significant variations in the moves and steps. Another proposals study was conducted by Yin (2016) who identified rhetorical structures used by Singaporean researchers in terms of their communicative purposes, institutional expectations and disciplinary culture. Dang (2023) scrutinized the employment of metadiscourse in RPs written by Vietnamese postgraduate students in Applied Linguistics and found that self-mentions were most frequent in the corpus, while attitude markers were least utilized. Nuttall (2021) explored grant proposal from a different angle focusing on lexical frame use. Adopting the frame-first approach to analyse frequently occurring four-word lexical frames, the study revealed that (1) many of the frames exhibit fixed variability in addition to being either unpredictable or highly unpredictable, and (2) the majority of frames are functionally referential while there are very few discourse organizing frames.

The term metadiscourse was created by Harris in 1959 to indicate the ways in which the writer/speaker conduces the receiver’s perception of a text through the use of language. Since then, the concept of metadiscourse has been further developed in a number of studies. Vande Kopple (1985) and Crismore and Farnsworth (1990), for instance, included into the metadiscourse category rhetorical features such as hedges, connectives and commentaries used to influence the reader’s perception of the text.Hyland (2005a) defined metadiscourse as «self-reflective expressions used to negotiate interactional meanings in a text, assisting the writer (or speaker) to express a viewpoint and engage with readers as members of a particular community» (Hyland, 2005a, p. 37). These linguistic resources help «relate a text to its context by assisting readers to connect, organise, and interpret material in a way preferred by the writer and with regard to the understandings and values of a particular discourse community» (Hyland & Tse, 2004, p. 157).

A wide range of taxonomies of metadiscourse has been developed in the last decades (Beauvais, 1989; Hyland, 2005b; Vande Kopple 1985). The present study adopted a more recent taxonomy proposed by Hyland (2005b) that eliminated overlapping functions and metadiscourse categories identified by formal, rather than functional, characteristics. The taxonomy encompasses the interactional aspects of discourse, using the criteria of external and internal relations and seems to more clearly identify rhetorical functions and reduce overlapping functions. Hyland distinguished between two types of metadiscourse - interactive and interactional. While the interactive metadiscourse helps organize a text coherently by linking sentences to each other so that the reader can understand it better, interactional metadiscourse is employed to interact with the reader, explicitly convey views and attitudes, and involve the audience by allowing them to respond to the unfolding text and anticipating their objections (Hyland, 2005a, p. 49-50). Interactional metadiscourse involves five elements - hedging, boosting, attitude stance, self-mention, and engagement - that «help realize the rational, credible, and affective appeals that contribute to the persuasiveness of a text» (Hyland, 1998, p. 226). Since the focus of the current study is boosting and hedging, consider these categories more closely.

Hedges and boosters are described as «communicative strategies for recognizing contingency and indicating the room the writer is willing to offer for negotiation» (Hyland, 2005b, p. 144). Hedges are employed to acknowledge alternative viewpoints and withhold commitment to the proposition. According to Hyland (2005b, p. 145), using a hedge the author expresses willingness to negotiate a claim thereby reducing commitment and conveying respect for alternative views. In contrast, boosters are used to «suppress alternatives, presenting the proposition with conviction while marking involvement, solidarity and engagement with readers» (Hyland, 2005b, p. 145). They are also used to strengthen an argument by suggesting the reader draw the same conclusions as the writer. As Hyland (2005a) claims, «the balance of hedges and boosters in a text thus indicates to what extent the writer is willing to entertain alternatives and so plays an important role in conveying commitment to text content and respect for readers» (p. 53).

A large number of works have demonstrated that «academic discourse is representative of shared cultural understandings about how discourses are constructed, through which communicative choices» (Khoutyz, 2015, p. 135). Kaplan (1966), for example, explained differences in academic writing in terms of differences in cultural backgrounds of the authors and claimed that writing style is not universal and varies «from culture to culture and even from time to time within a given culture» (p. 2). Kaplan distinguished between four types of writing such as linear paragraph development, a series of parallel constructions, cyclic paragraph development, and digression. Kaplan (1966) argued that it is apparent that paragraph organisation written in English by a L2 writer «will carry the dominant imprint of that individual’s culturally-coded orientation to the phenomenological world in which he lives and which is bound to interpret largely through the avenues available to him in his native language» (p. 1).

One more study on writing styles was conducted by Galtung, who also distinguished between four types - Gallic (prototype: the French), Teutonic (prototype: the Germans), Saxonic (prototype: the English and the Americans), and the Nipponic (prototype: the Japanese and Far East Asians). Here are some features of these styles that are relevant for the present study.

While the Saxonic style fosters and encourages debate and discourse, the dispersion or diversity of opinion in Gallic and Teuton cultures is likely to be smaller, and in the Nipponic setting, people are not very skilful at debating.

When Galtung compares these four cultures, he notices that Eastern Europe found itself under the influence of the Teutonic style, «partly because of general cultural influence through centuries, partly because of the influence of a key Teutonic thinker: Karl Marx» (Galtung, 1981, p. 820). The development of Russian and Ukrainian scholarship features, therefore, as an illustration of Teutonic influence. In addition, these languages are typologically disposed to tolerate similar textual patterns as those that are typical of German. The Asian countries are considered to be under the influence of the Nipponic style affected by Confucian, Taoist, and Buddhist philosophical precepts with Japan seen as a centre of this type of culture.

Another taxonomy of writing styles was developed by Hinds (1987). It is based on reader- and writer-responsible cultures: in contrast to writer-responsible writing style, reader-responsible one is characterized by an indirect form of argumentation with a less-assertive conclusion and requires a great deal of inferential work on the part of the reader. According to Hinds (1987), in writer-responsible culture (Anglophone countries), it is the writer who is primarily responsible for effective communication, while in reader-responsible one (e.g., Japan or German), it is the responsibility of the reader to understand what the writer had intended to say as they place the burden on the reader to discern a meaning.

Based on these theories, Duszak (1994) revealed differences in metadiscourse patterns used in academic prose by native and non-native (Polish) speakers of English and explained them in terms of Galtung’s (1981) model. Kim and Lim’s (2013) study based on Hinds’ (1987) theory proved that Asian rhetorical traditions favour the reader-responsible attitude when «the writer controls the level of personality in a text to establish a more distant relationship between author, text and reader» (p. 140). Similarly, Qi and Liu (2007) argue that «compared to apparent formal links adopted in English, Chinese, in most cases, may have covert structural clues in the author’s mind, requiring the reader to explore cohesion in the discourse» (p. 148). In the same vein, Park and Kim (2008), who explored the communication styles of Asian and American communities, revealed that Asian speakers prefer a less open communication style due to their higher level of emotional self-control and tend to suppress emotions by avoiding personal opinions or attitudes.

Thus, previous studies have identified a significant intercultural variation in the rhetorical preferences of academic writers, indicating that writing is a cultural object shaped by culture. As Hyland (2005a) put it, the rhetorical style and the degree of writer’s rhetorical assertiveness are closely associated with cultural conventions.

The present study aims to explore writing style-dependent differences in the employment of hedges and boosters seeking to answer the following questions:

What categories of metadiscourse - hedges or boosters - do Asian and Slavic novice writers mainly opt for in their RPs?

To answer the research questions, a mixed methods research design was adopted. The following subsection will provide details concerning participants, instruments, procedure, and methods used in the present study.

116 RPs were selected and divided into two parts. The number of tokens in each subcorpus was 118 123 and 124 569, which makes 243 692 tokens altogether. Due to the small corpus, the present study does not seek to allow generalizations about the employment of metadiscourse markers in RPs by postgraduate students exposed to different writing styles.

The first subcorpus consisted of 58 English RPs written by postgraduate students from Vietnam and China. The second sub-corpus also consisted of 58 English RPs written by Russian native students. The examples quoted in this paper are coded by indicating the number of the subcorpus: subcorpus (SC) 1- for the Asian subcorpus and subcorpus (SC) 2 - for the Slavic subcorpus.

To investigate hedges and boosters in RPs written by the students exposed to different writing styles thus achieving the goal of the study, the methods of quantitative and contextual analysis were adopted.

Hyland and Zou’s (2021) typology of boosting markers was adopted as the initial model for revealing boosters. Hedging devices found in the corpus were analyzed using the model by the same authors (Hyland & Zou, 2021) who identified three ways of conveying respect for colleagues’ views. The taxonomies adopted in the current study are presented in Table 1.

Table 1.

Categories	Types	Function
Hedges	Plausibility hedges	signal that a claim is based on assumptions
	Downtoners	mitigate the intensity of a statement
	Rounders	indicate an approximation
Boosters	Certainty markers	indicate the writer’s epistemic conviction
	Extremity markers	emphasize the upper edge of a continuum
	Intensity markers	amplify the emotive strength of a statement

Frequencies of each category of boosters and hedges presented in Table 1 were used to determine their dominance in each subcorpus. Afterward, contextual analysis was adopted to elaborate on the results from the quantitative analysis. This study also employed a comparative approach as it sought the realization of hedging and boosting in RPs written by the postgraduate students exposed to different writing styles.

The analysis process went through several steps. Quantitative analysis was applied to all cases of hedges and boosters in the two subcorpora so as to analyze the socio-pragmatic context in which linguistic means are used to identify whether they act as metadiscourse markers.

First, hedges and boosters were identified manually in 116 RPs. Second, the markers found in the corpus were manually analyzed in context. Since the sizes of the two sub-corpora were not equal, the raw frequencies of occurrence of stance markers and normalized frequencies of the number of occurrences per 1000 words were determined. The occurrences were processed automatically with a freeware, multiplatform tool for carrying out corpus linguistics research AntConc 3.4 Windows, an advanced text analysis application which provides details about the text and can ensure the accuracy of research results. Then, all the examples found in the corpus were analyzed in context to describe the rhetorical functions of hedges and boosters.

Hedges downplay writers’ commitments to propositional content, modifying its relevance or certainty and helping to acknowledge alternative viewpoints. They withhold commitment to the presented proposition and are used to steer the reader to the conclusion or reasoning of the writer’s choice (Boginskaya, 2022c).

The analysis revealed that both Asian and Slavic writers employed all three types of hedges. Plausibility hedges more frequently used by Slavic students (53.2 vs 197.9 per 1000 words) serve to protect the author from having to take full responsibility for the propositional content and to recognize the limitations of claims. Here is an example of the plausibility hedge used by the Asian student to indicate that the statement is based on an assumption rather than facts and to implicate that the author is uncertain about the proposition.

Downtoners help protect the writer against inaccuracy of research results, as illustrated in example 2.

In example 3, the downtoner usually might convey certain qualification with regard to the degree of accuracy of the conclusion, demonstrating that the statement might be inaccurate.

The analysis revealed that downtoners were also more frequently used by Slavic students mainly to protect themselves from inaccuracies.

Rounders indicating an approximation - were rather scarce in both sub-corpora. Here is an example from SC1:

By making the number a little fuzzy, the adverb employed as a rounder expresses approximation, thereby making the claim less persuasive. In addition, the authors can use this type of hedging when they do not know precise terms or numbers or they are no relevant.

In contrast to hedges, boosters function by «presenting the proposition with conviction while marking involvement, solidarity and engagement with readers» (Hyland, 2005a, p. 145). An analysis has revealed the higher frequency of these devices per 1000 words in SC1, which indicates that the Asian writers tended to occupy a far stronger stance and were more keen to express their convictions and highlight the significance of their work:

The booster used in the above example helps remove any doubts about the claim closing down potential opposition.

In the two sub-corpora, boosters differed both numerically and functionally. While certainty boosters were used more frequently by Asian writers, in the Slavic sub-corpus intensity boosters prevailed. Extremity boosters were rarely employed by both groups.

Certainty boosters were used to indicate the writer’s epistemic conviction, to claim the accuracy of research results, to emphasize the importance of the study, and to exclude alternative views from readers as follows:

In the above examples, the authors anticipate possible responses from the reader but choose to prevent them. The boosting verbs to confirm and to demonstrate and the boosting noun findings are used to express authors’ certainty in research results obtained or claims presented.

The adverb evidently is used here to signal accepted truth - that is, it downplays the author’s involvement by implying that the claim is one that is already generally accepted in chemical engineering.

Intensity boosters function by amplifying the emotive strength of a statement. In contrast to certainty boosters, they add affective color to claims rather than concern epistemic assurance (Hyland & Zou, 2021).

These intensity boosters function by enhancing persuasion through an involved attitude. Regarding the extremity boosters, they «emphasize the upper edge of a continuum» (Hyland & Zou, 2021, p. 8), as in here:

By upgrading the propositions, the writers emphasize the level of parts’ stability, as seen in example 11, and the significance of features, as seen in example 12, without the need for elaboration.