Hedging practices in soft science research articles: A corpus-based analysis of Indonesian authors

Abstract

Hedging devices play a crucial role in mitigating claims and reducing directness in academic writing. Their utilisation in scholarly discourse facilitates tentativeness and fosters effective communication. The present study investigates the use of hedging devices by Indonesian researchers in the soft sciences. The dataset comprises 2,511,423 words from 400 research articles authored by Indonesian scholars. The results demonstrate that Indonesian researchers in the field of soft sciences utilised hedges at a rate of 103.77 words per 10,000 words, which is comparatively lower when contrasted with the usage of hedges observed in articles published in internationally renowned journals authored by both native and non-native English speakers. Additionally, modal auxiliary verbs were identified as the most commonly employed category of hedging devices, with the modal verb can being the predominant hedging item across all types of hedges. The findings contribute to understanding hedging practices among non-native English speakers, particularly in the Indonesian academic context. This study concludes by suggesting a potential area for improvement in academic writing among Indonesian researchers, especially in the humanities and social sciences.

Keywords:

• corpus linguistics
• hedging devices
• English for academic purposes
• academic discourse
• education
• Indonesia

1. Introduction

Hedging is a linguistic device used by a writer/speaker to avoid a complete commitment to a proposition (Alghazo et al., 2021). Writers or speakers employ hedging techniques to moderate the strength of a statement, thereby avoiding excessive directness in their assertions, and “allowing information to be presented as an opinion rather than accredited fact” (Hyland, 2005, p. 178). The term “hedge” can be traced back to the seminal work of Lakoff (1973), in which he elucidated its function as a linguistic device that serves to introduce a certain degree of vagueness, caution, and indecisiveness. Within the realm of academic writing, hedges serve not only to modulate the level of commitment to a proposition but also fulfil various significant pragmatic functions. These include facilitating the creation of a dialogic space, acknowledging subjectivity, mitigating potentially face-threatening criticisms (Chen & Zhang, 2017), and maintaining objectivity (Swales & Feak, 1994).

Additionally, Hyland (1996) stated that the use of hedging is significant in academic discourse as it enables tentativeness and opens possibility for communication. The perception of academic discourse has evolved from the notion that it solely conveys objective and impersonal truths to recognising it as a form of persuasive discourse, acknowledging the potential for audience interaction and engagement with the writer (Hyland, 2005). The judicious application of hedging strategies enables academic article authors to effectively navigate their relationships with their own propositions and arguments, while also contributing to the establishment of power dynamics between authors, readers, and reviewers (Mu et al., 2015).

Moreover, the proficient and appropriate utilisation of hedging devices is also paramount due to the integral role hedging plays in a writer’s pragmatic competence (Chen & Zhang, 2017, Fraser, 2010). The pragmatic competence refers to the writer’s capacity to utilise language effectively within dynamic contexts, enabling them to produce an academic text that aligns with the rhetorical norms and expectations of a specific discourse community (Chen & Zhang, 2017). Hence, writers should possess the knowledge and skill to employ appropriate hedging devices when composing academic articles since the absence or misuse of hedging can result in pragmatic failure in written communication (Chen & Zhang, 2017). However, despite its significance as a prominent feature in academic writing, the underuse of hedging is frequently observed among non-native users of English (Almakrob, 2020; Gomaa, 2019; Johansen, 2020; Thabet, 2018).

Numerous studies have been undertaken to investigate the utilisation of English hedges by various L1s in academic writing, including student papers, dissertations, and journal articles (e.g., Almakrob, 2020; Gomaa, 2019; Thuy, 2018; Smirnova & Strinyuk, 2020; Takimoto, 2015). The majority of these studies have focused on comparing the usage of hedges between native and non-native English speakers. For examples, Almakrob (2020) conducted a study examining the use of hedges in PhD dissertations written by L1 Arabic, while Shafqat et al. (2022) analysed the employment of hedges by Pakistani PhD students. Both studies revealed findings indicating a limited use of hedging in their written discourse. Regarding the utilisation of hedging in research articles, Farrokhi and Emami (2008) conducted a comparative analysis on the deployment of hedges by Iranian writers and native English writers in the domains of applied linguistics and engineering research articles, while Yagız and Demir (2014) examined the implementation of hedging strategies by Turkish writers and their native counterparts in ELT (English Language Teaching) journal articles. These studies also highlight a significant disparity in the frequency of hedging devices employed by native and non-native English writers, with native writers showing a higher usage of hedges compared to their non-native counterparts. In a similar vein, Tran and Tang (2022) examined hedges exercised by L1 Vietnamese and L1 English writers in English applied linguistics research articles. Nevertheless, their research revealed that both Vietnamese and English writers employed hedges at a comparable frequency. This suggests that factors such as different first language backgrounds and academic disciplines may play a role in shaping the employment of hedging strategies in research articles. This is also emphasised by Takimoto (2015) who found that professional writers in the field of social sciences tend to apply a greater degree of hedging in their scientific articles compared to those in the natural sciences. In this regard, there is a need for research investigating the application of hedges across diverse disciplines and language backgrounds, including the Indonesian context.

While there have been some studies conducted on the use of hedges by L1 Indonesians, such research remains limited. Notably, Asfina et al. (2018) investigated the utilisation of English hedges by Indonesian ELT students in both written and spoken discourses. Additionally, Sanjaya (2015) and Sanjaya et al. (2015) examined whether there are significant differences in the frequency of hedges used in English and Indonesian applied linguistic research articles, with native speakers from each respective language serving as the authors. The latter study had a cross-linguistic approach, comparing the usage of English hedging devices by native English writers and Indonesian hedging devices by L1 Indonesian speakers. Nonetheless, further research is necessary to comprehensively explore the use of hedges by L1 Indonesians in diverse contexts and disciplines. To address this gap, the present study endeavours to analyse the use and function of English hedging devices by Indonesian researchers in a corpus consisting of soft sciences journal articles published between 2019 and 2022. The choice of investigating soft science articles in the Indonesian context stems from the notable gap in existing literature, wherein no prior studies have comprehensively examined the utilisation of hedges in this specific domain, underscoring the study’s aim to contribute novel insights into the linguistic and rhetorical strategies employed by L1 Indonesian researchers in negotiating uncertainty and cautiousness within their academic writing. Moreover, the selection of the data collection timeframe between 2019 and 2022 is driven by the objective of reflecting current language use, ensuring that the study captures the most up-to-date practices of Indonesian researchers in employing hedges within soft science articles.

The findings of this study may provide valuable insights into the specific patterns and functions of hedging devices employed by Indonesian researchers in soft sciences journal articles. These insights can have practical implications for the teaching and instruction of hedging in English for Academic Purposes (EAP) classrooms or academic writing programs in Indonesia. By understanding the specific linguistic strategies used in hedging, educators can enhance the effectiveness of teaching hedging to Indonesian students, thereby improving their academic writing skills and facilitating their successful participation in academic discourse.

2. Literature review

2.1. Taxonomy of hedging devices

There have been various taxonomies used for the classification and categorisation of hedging (e.g., Crompton, 1997; Hyland, 1996, 1998; Salager-Meyer, 1994). Salager-Meyer (1994, pp. 154–155), for example, developed a formal-functional classification of hedging based on a contextual analysis of written medical English discourse. This taxonomy classifies hedging devices into five distinct categories including “shields” (e.g., seem, appear), “approximators” (e.g., usually, somewhat), “expressions of authors’ personal doubt and direct involvement” (e.g., I believe, to our knowledge), “emotionally-charged intensifiers” (e.g., surprisingly, extremely difficult), and “compound hedges” (e.g., it could be suggested, it would seem likely that). This taxonomy, however, has faced criticism due to the presence of apparent overlaps between the defined categories (Chen & Zhang, 2017). Hyland (1998) and Crompton (1997), on the other hand, seem to have a similar taxonomy that is mainly based on the form/structure of hedges in academic discourse. A slight difference lies in the fact that Crompton’s taxonomy is confined to a selection of “items of language” or lexical items, whereas Hyland’s taxonomy include lexical hedges and strategic-based hedges. This study, however, employed Hyland’s (1996, 1998) categorisation specifically focusing on lexical hedges since they are the primary hedging devices employed by Anglophone writers (Hyland, 1994) and are likely to be recognised as hedges within the academic discourse community (Chen & Zhang, 2017; Varttala, 1999). Hyland (1996, 1998) divides lexical hedges into five major forms: modal verbs, epistemic lexical verbs, epistemic adjectives, epistemic adverbs, and epistemic nouns. Table 1 shows examples of lexical devices in each category.

Table 1. Examples of hedges

Types of Hedges	Examples
Modal Verbs	may (not), might (not), could (not), can (not), would (not), and so forth
Lexical Verbs	indicate, argue, suggest, tend, and so forth.
Adjectives	possible, likely, probable, plausible, and so forth
Adverbs	often, usually, rather, almost, and so forth
Nouns	Tendency, possibility, assumption, implication, and so forth

2.2. Functions of hedges

Previous scholars have extensively examined the functions of hedging and have put forth various taxonomies to describe these functions. Among these models, the ones proposed by Salager-Meyer (1994) and Hyland (1998) are widely recognised and commonly employed in numerous studies of hedges (Thuy, 2018). The present study utilised Hyland’s (1996, 1998) taxonomy to investigate the functions of hedges found in the corpus, following Thuy (2018), and Wang and Tatiana (2016).

According to Hyland (1996), hedges have two major pragmatic functions, which are content-motivated and reader-motivated. The content-oriented hedges are further divided into two types, namely accuracy-based and writer-based hedges. Accuracy-based hedges are used by writers when they try to express propositions with enhanced precision and caution, particularly in domains where interpretations may be subject to change (Hyland, 1996). They serve to differentiate between what is known as factual or actual and what is inferred or speculative. By using these hedges, writers indicate that a proposition is based on their plausible reasoning rather than on established or reliable facts (Hyland, 1996). This type of function is canonically realised by epistemic modal verbs, epistemic adverbs, and epistemic adjectives which signify limited knowledge of the writers. Additionally, writer-based hedges enable writers to avoid full personal responsibility for their statements and mitigate the potential consequences that may arise from making bald propositions (Hyland, 1996). In other words, this allows writers to maintain a level of flexibility and caution in their language, acknowledging the hypothetical nature of certain assertions while protecting themselves from potential criticism or backlash. This function is commonly realised by evaluative epistemic lexical verbs, typically in passive voice and impersonal expressions.

Furthermore, reader-oriented hedges “contribute to developing a relationship with readers by addressing the need for deference and cooperation in gaining the ratification of claims” (Hyland, 1996, p. 257), which function as politeness signals (Lee-Wong, 2005). These hedges are used to mark statements as provisional or tentative, recognising that the readers have the ability to critically evaluate and form their own judgments on the matter (Hyland, 1996). This can be realised by deductive epistemic lexical verbs involving the use of personal attributions. It should be noted, however, hedges can also exhibit polypragmatic characteristics as they may serve multiple functions simultaneously.

3. Method

The data used in this study consisted of a corpus comprised of research articles written by Indonesian authors in the field of soft sciences. The data collection process involved the use of purposive sampling, whereby specific criteria are established to guide the selection of articles for inclusion in the corpus. First, the journals, as the data source, should be indexed by Scopus or accredited by Indonesian Ministry of Research, Technology and High Education in SINTA 1. It is significant to note that all of the chosen journals are Indonesian-based since they “tend to reflect the language produced by Indonesian authors in Indonesian contexts for international audiences” (Fajri et al., 2020). Furthermore, the chosen articles ought to be authored by Indonesian writers within the timeframe of 2019 to 2022, as this will effectively demonstrate the contemporary patterns pertaining to the utilisation of linguistic hedges.

To ensure that the writers of the articles are Indonesians, the present study followed the method by Fajri et al. (2020), which adopted Wood’s (2001) framework. They define native Indonesian writers as those whose first and last names are typical Indonesian names and those who are affiliated with Indonesian universities. In the case of multiple-authored articles, only those in which all authors were of Indonesian nationality were included in the study. Additionally, journal articles that focused on English studies, EFL (English as a Foreign Language), and linguistics were excluded since extensive research on hedges in these fields has already been conducted. This resulted in 400 papers collected from 10 journals in accordance with the criteria. The disciplines of the selected journals include Psychology, Law, Economics, Area Studies, Arts-Humanities, Social-Political Sciences, and Philosophy. All the collected articles were first cleaned from the unnecessary words, such as authors’ names and affiliations, references, pages, tables, figures, and the name of the journal, since they naturally do not contain hedging devices used by the authors. Then, the resulted articles were converted into text files so that they could be processed by the corpus tool. In total, the corpus comprised 2,511,423 tokens. Table 2 presents an overview of the included journals in the study, along with the corresponding number of articles and tokens associated with each journal.

Table 2. Distribution of the corpus

No.	SINTA 1 Indexed Journals	Number of Articles	Academic Disciplines	Number of Tokens
1	Psikohumaniora: Jurnal Penelitian Psikologi	57	Psychology	267,136
2	Journal of ASEAN Studies	28	Area Studies	150,386
3	Padjajaran Jurnal Ilmu Hukum	51	Law	465,135
4	Ijtihad: Jurnal Wacana Hukum Islam dan Kemanusiaan	36	Law	208,098
5	Jurnal Ilmu Sosial dan Ilmu Politik	53	Social-Political Sciences	342,572
6	Harmonia: Journal of Arts, Research, and Education	53	Arts and Humanities	265,163
7	Gadjah Mada International Journal of Business	29	Economics	214,527
8	Economic Journal of Emerging Markets	17	Economics	82,929
9	Paramita: Historical Studies Journal	56	History	350,404
10	Teosofi: Jurnal Tasawuf dan Pemikiran Islamic	20	Philosophy	165,073
	TOTAL	400		2,511,423

The corpus software LancsBox (Brezina et al., 2015) was employed to retrieve the hedges. The technique of data analysis of the present study followed the framework proposed by Hyland (1996), which includes the surface-level quantitative analysis and polypragmatic function analysis. For the surface-level quantitative study, frequency analysis was employed to identify the hedging devices present in the corpus. The lemma option was also employed to account for the potential ambiguity of words with multiple parts of speech. By using lemmatization, the findings are aligned with the base form of each word, ensuring greater accuracy in the categorisation of hedges based on Hyland’s (1998) models. Moreover, in order to gain insights into the functions of the identified hedging devices within Hyland’s (1996) poly-pragmatic model, the study employed the concordance tool, which facilitates a qualitative analysis by displaying the occurrences of each hedge along with its co-textual information simultaneously.

4. Results

4.1. Overall distribution of hedging devices

Table 3 presents the overall distribution of hedging devices. The raw frequency of all hedging devices in the corpus were 26,097, or 103.91 per 10,000 words. This number is relatively low, especially for argumentative-driven disciplines, which has been shown to contain more hedging devices than hard sciences (Hyland, 2004; Hyland & Jiang, 2016; Wang, 2022). For example, Hyland and Jiang (2016) found that the frequency of hedges in sociology research articles published in 2015 were 148.7 per 10,000 words, while Babaii et al. (2015) demonstrated that the occurrences of hedging devices in psychology and sociology scientific articles were around 160.7 and 126.1 respectively.

Table 3. Overall distribution of hedging devices

No	Grammatical Form	Raw Frequency	Relative Frequency (per 10,000 words)
1	Modal verbs	15253	6.73
2	Lexical verbs	5150	2.51
3	Adverbs	3172	12.63
4	Nouns	1276	5.08
5	Adjectives	1246	4.96
	TOTAL	26097	103.91

In addition, modal verbs were the most frequent used form of hedging devices, occurring 15,253 times, with can as the most recurrent hedging item across all types of hedges. Following modal verbs, lexical verbs were the next most commonly employed hedging form, with a total frequency of 5,150 occurrences. Adverbs were used as hedging devices 3,172 times, while nouns and adjectives were employed 1,276 and 1,246 times respectively. This finding is congruent with Wang and Tatiana (2016) who also found that modal auxiliaries were the most used hedging devices. The following sections provide a detailed description in the use of each type of hedges.

4.2. Modal auxiliary verbs as hedging devices

Table 4 provides an overview of the modal verbs identified as hedging devices in the corpus. The analysis shows a total of 15,253 occurrences of modal auxiliary verbs. The most frequently used modal verb was can, which appeared 8,044 times, accounting for 52.73% of the total number of modal verbs. Following can were the modal verbs should, may, could, and would, each of which had at least 1,000 occurrences in the corpus.

Table 4. Frequency of modal verbs as hedging devices

No	Item	Raw Frequency	Relative Frequency (per 10,000 words)	%
1	Can	8,044	32.05	52.73%
2	Should (not)	1,679	6.69	11%
3	May (not)	1,582	6.30	10.37%
4	Could (not)	1,476	5.88	9.67%
5	Would (not)	1,064	4.23	6.97%
6	Cannot	842	3.35	5.52%
7	Might (not)	307	1.22	2.01%
8	Shall	248	.98	1.62%
9	Ought	11	.07	0.07%
	TOTAL	15,253	6.74	100%

The prevalence of the modal auxiliary verb can among Indonesian writers in the corpus aligns with the findings of Abdollahzadeh (2019) and Demir (2018), who examined the use of hedges by Iranian and Turkish writers in applied linguistics and language education journals respectively. However, this differs from the use of hedges by native writers that tend to prefer the modal verb may as a hedging device, as found in the previous studies (e.g., Hyland, 1998; Orta, 2010; Thuy, 2018). Also, the modal verb would is recurrently used by native journal article authors (Hyland, 1998; Thuy, 2018), but the present study reveals that would ranks as the fifth most frequent modal verb in the corpus. The contrasting usage patterns between Indonesian writers and native English speakers may reflect cultural and linguistic differences in their approach to hedging. It seems that native English-speaking authors frequently employ may or would as a hedge to introduce a level of uncertainty and mitigate potential contradictions between themselves and the readers (Thuy, 2018). This choice allows for greater flexibility in interpretation. On the other hand, Indonesian writers’ preference for can might indicate that they focus more on expressing possibility or capability, showing more certainty. Regarding the least frequently used modal verb, similar to the findings of Hyland’s (1996) work, the current study also demonstrates that ought was rarely employed by Indonesian authors. This consistency suggests a shared pattern in the limited usage of ought as a hedging device in academic writing.

The epistemic modal verbs were predominantly used as accuracy-based hedges. Writers employ accuracy-based hedging devices with the intention of presenting their written work in an objective and precise manner. These devices serve to address any potential inaccuracies by providing a sense of acceptability and allowing for the specification of the level of accuracy. Examples of accuracy-based hedges through the use of epistemic modal verbs are demonstrated in Excerpt 1 and 2

1. Whereas, horizontal spillovers in a long-term may arise through demonstration effect, labor mobility and competitions.

2. Integrity should be seen as a dynamic process in the formation of self-identity.

4.3. Lexical verbs as hedging devices

Epistemic lexical verbs serve as hedging devices to mitigate commitment or assertiveness in writing. Table 5 displays the utilization of lexical verbs as hedges identified within the sample data.

Table 5. Frequency of lexical verbs as hedging devices

No	Item	Raw Frequency	Relative Frequency (per 10,000 words)	%
1	Indicate	946	3.76	18.36
2	Tend	806	3.21	15.65
3	Argue	740	2.94	14.36
4	Suggest	468	1.86	9.08
5	Report	464	1.84	9
6	Seem	431	1.71	8.36
7	Appear	426	1.69	8.27
8	Propose	394	1.57	7.65
9	Predict	271	1.07	5.26
10	Assume	204	.81	3.96
	TOTAL	5,150	2.46	100

It can be seen that the lexical verbs employed to express hedging in the corpus amount to 5,150 instances, corresponding to a frequency of 20.46 occurrences per 10,000 words. The verb indicate emerged as the most frequently used lexical verb, with 946 occurrences, aligning with the findings reported by Hyland (1996). Additional notable lexical verbs include tend and argue, which exhibit 806 and 740 occurrences, respectively. However, in contrast to modal verbs which were dominated by the use of can, the distribution of lexical verbs in the table appears to be relatively even, indicating more comparable levels of usage.

This finding, nevertheless, presents a slight difference from previous research conducted by Abdollahzadeh (2019) and Wang and Tatiana (2016), where the verb suggest was identified as the most frequently utilised form of hedging. This difference could potentially be attributed to the influence of different disciplines, as the previous studies focused on linguistics and applied linguistics research articles, while the current study has included a broader range of disciplines. Disciplinary variations can impact the choice and frequency of hedging devices used by researchers in their respective fields of study (Vold, 2006; Wang, 2022). It is worth noting that although indicate may not be the most prevalent hedging verb, it remains among the highly utilised lexical verbs for expressing mitigation, in the aforementioned studies by Abdollahzadeh (2019) and Wang and Tatiana (2016).

Furthermore, the analysis revealed that epistemic lexical verbs were predominantly employed as writer-based and reader-motivated hedges. This can be observed in Excerpts 3 and 4, where writer-based hedges were utilised using the epistemic lexical verbs indicate and tend, respectively.

• (3) The results of this study also indicated that there is a significant effect of perceived happiness on the quality of work-life in employees who experience job-skills mismatch.

• (4) The economic and political interests of the city tend to be impartial to heritage buildings.

Additionally, lexical verbs were observed to function as reader-oriented hedges when preceded by the first-person pronouns or noun phrases referring to the authors (e.g., I, we, the writers). This particular hedging strategy is employed when the author is unable to acquire sufficient evidence to confidently make deductions regarding the issue being investigated (Loi & Lim, 2019). Excerpt 5 and 6 provides examples of the use of lexical verbs as reader-oriented hedges.

• (5) In this case, we argue that AUMA and FKM tend to take side with the Wahhabism in its purification and intolerant behavior but not in its religious practice.

• (6) Therefore, we suggest that a transformation of the internet cafe creates a change in consumption patterns.

4.4. Adverbs as hedging devices

Epistemic adverbs can be utilised to hedge, conveying a sense of indefiniteness or imprecision regarding the information presented (Wang, 2022). Table 6 provides an overview of various adverbs used for hedging, along with their respective frequencies.

Table 6. Frequency of adverbs as hedging devices

No	Item	Raw Frequency	Relative Frequency (per 10,000 words)	%
1	Often	772	3.07	24.17
2	Usually	370	1.47	11.58
3	Rather	369	1.47	11.55
4	Generally	341	1.35	10.67
5	Almost	285	1.13	8.92
6	Relatively	272	1.08	8.51
7	About	269	1.07	8.42
8	Quite	216	.86	6.76
9	Around	128	.51	4
10	likely	70	.27	2.19
11	Perhaps	51	.2	1.59
12	Probably	51	.2	1.59
	TOTAL	3194	12.68	100

There were 3,194 occurrences of adverbs as hedging devices in the corpus. The adverb often held the highest frequency, occurring 772 times. Following closely were the adverbs usually, rather, and generally, comprising 11.58%, 11.55%, and 10.67% of all adverbs, respectively. These findings echo the studies conducted by Takimoto (2015) and Wang and Tatiana (2016), which found that often was the most commonly used hedging adverb. This consistency across different studies may suggest the significance of adverbs, particularly often, in expressing hedging in academic writing, especially in humanities and social sciences.

Regarding the hedging functions of epistemic adverbs, they were predominantly employed as accuracy-based hedges. Specifically, epistemic adverbs were commonly utilised as downtoners. This can be exemplified in Excerpts 7 and 8, which showcase instances of epistemic adverbs found within the corpus.

• (7) Efforts to bring about justice often conflict with political considerations aimed at maintaining power.

• (8) The seizing states are usually reluctant to use such extensive powers by prosecuting and bringing arrested pirates to their courts.

4.5. Adjectives as hedging devices

Epistemic adjectives can also serve as tools for expressing mitigation. They can be employed to convey information that is considered uncertain or imprecise, as noted by Wang and Tatiana (2016). Table 7 provides the distribution of adjectives used as hedging devices within the corpus.

Table 7. Frequency of adjectives as hedging devices

No	Item	Raw Frequency	Relative Frequency (per 10,000 words)	%
1	Most	542	2.15	43.49
2	Possible	351	1.39	28.17
3	Likely	278	1.10	22.31
4	Rare	31	.12	2.48
5	Unlikely	26	.10	2.08
6	Plausible	12	.04	0.96
7	Probable	6	.02	0.48
	TOTAL	1246	4.92	100

As shown in Table 7, the total number of hedging adjectives found in the corpus was 1,246, or 4.92 per 10,000 words. Most was the most frequent adjective used as a hedging device, accounting for 43.49% of all adjectives in the corpus. It is applied as an epistemic quantification which makes the proposition less absolute (Takimoto, 2015). This finding aligns with the studies conducted by Hyland (1996) and Wang and Tatiana (2016), which similarly identified the adjectives most, possible, and (un)likely as among the most commonly used adjectives for expressing hedging. Additionally, epistemic adjectives were mostly used as accuracy-based hedges as shown in the extracts 9 and 10.

• (9) Because most multinational companies in Indonesia are export-oriented and generally do not supply to Indonesian customers, the focus of this study is only on technology transfer through backward spillover.

• (10) The trend is likely to happen because Malaysia’s export in the selected markets did not grow as fast as the growth of the world.

4.6. Nouns as hedging devices

Table 8 presents a detailed breakdown of the frequencies of nouns used as hedging devices. The finding reveals that these hedging nouns were frequently derived from lexical verbs and adjectives, which is consistent with the observations made by Hyland (1996).

Table 8. Frequency of nouns as hedging devices

No	Item	Raw Frequency	Relative Frequency (per 10,000 words)	%
1	Tendency	341	1.35	26.72
2	Implication	317	1.26	24.84
3	Possibility	227	.9	17.78
4	Assumption	156	.62	12.22
5	Probability	150	.59	11.75
6	Prediction	46	.18	3.6
7	Doubt	39	.15	3.05
	TOTAL	1276	5.05	100

The noun tendency is the most frequently used hedging noun in the corpus, appearing 341 times. Furthermore, the other four most frequently occurring hedging nouns are implication (317 times), possibility (227), assumption (156), and probability (150). This finding further supports Hyland’s (1996) assertion that nouns used for hedging purposes often originate from verbs and adjectives, as evidenced by the derivations observed in these examples. Extract 11 and 12 show examples of epistemic nouns utilised as a form of mitigation.

• (11) Firstly, training has a greater tendency to make a salesperson motivated and enthusiastic.

• (12) There is also a possibility that it is the nature of resilience that makes an individual’s future time perspective increase.

5. Discussion

The present study focuses on examining the frequency of hedges usage in a corpus of soft sciences research articles authored by Indonesian writers. The findings indicate that Indonesian researchers in the soft sciences employed hedges at a rate of 103.77 words per 10,000 words. This rate is relatively low compared to the usage of hedges in articles published in internationally recognised journals written by both native and non-native English speakers, as indicated by prior studies (e.g., Hyland & Jiang, 2016; Wang, 2022). The relatively low usage of hedging devices by non-native English writers in journal articles may suggest a less flexible utilisation of hedges compared to native English speakers (Tran & Tang, 2022).

The present study also found that the most used hedging devices were modal auxiliary verbs, which is in line with previous research on the use of hedges by native or non-native English academic article writers (e.g., Abdollahzadeh, 2019; Tran & Tang, 2022). This may imply a consistency in the preference for modal verbs as a common strategy for hedging across different language backgrounds and academic contexts. However, the higher frequency of can used by Indonesian authors and other non-native English writes compared with their native counterparts may suggest the need for Indonesian writers to enhance the way they use modal verbs as hedging devices.

The findings of this study hold valuable implications when considering the context of articles published in journals categorized as SINTA 1 in Indonesia, which are esteemed as indicators of high-quality research. The observed relatively low frequency of hedges in the corpus of soft sciences research articles authored by Indonesian writers, when compared to usage patterns in internationally recognised journals as identified in prior studies, may suggest a nuanced perspective. It could be inferred that while these articles are categorised as high-quality, the usage of hedges in this particular corpus might indicate a potential area for enhancement in academic writing practices among Indonesian researchers, specifically in terms of expressing uncertainty, caution, and scholarly stance.

More specifically, these findings may have several implications for English language teaching and academic writing instruction in Indonesia. Firstly, the lower utilisation of hedges may indicate a need for English language teachers and instructors to focus on raising awareness and providing explicit instruction on the effective use of hedging devices in academic writing. Enhancing the understanding and application of hedges can contribute to the development of more nuanced and sophisticated writing skills, especially in argumentative-driven disciplines where the use of hedging devices tends to be more prevalent.

Secondly, the dominance of modal auxiliary verbs, particularly the frequent use of can, as the primary form of hedging device in the corpus highlights a specific pattern in the expression of uncertainty and mitigation by Indonesian writers. This preference for can may suggest a tendency to focus more on expressing possibility or capability rather than introducing a level of uncertainty. English language teachers and academic writing instructors should address this pattern and encourage a broader range of modal verbs as hedges to enhance the flexibility and precision in expressing different degrees of certainty and mitigation. Comparisons with native English speakers’ usage of modal verbs, such as the preference for may or would, can be used to highlight the cultural and linguistic differences in approaching hedging strategies.

Furthermore, the discipline-specific variations in the usage of hedging devices found in the study indicate the influence of academic disciplines on the choice and frequency of hedges. This implies that English language teaching and academic writing instruction should take into account the specific disciplinary contexts in which students are writing. Providing genre-specific guidance and teaching discipline-specific hedging practices can assist Indonesian researchers in effectively communicating uncertainty, mitigating potential contradictions, and maintaining the appropriate tone within their respective fields of study.

Lastly, the findings also shed light on the different functions of hedging devices. English language teachers and academic writing instructors can use these findings to illustrate the various functions of hedging devices and provide students with explicit instruction on how to employ them appropriately in different contexts. Academic writing instructors can also utilise a corpus as a reference or teaching materials. Native and English learner corpora might be consulted to gain insights regarding how native English speakers and learners use hedges in their writing. This will enable students to develop a deeper understanding of the rhetorical purposes and pragmatic implications of hedging in academic writing. In conclusion, addressing the relatively low frequency of hedges, promoting a wider range of modal verbs as hedging devices, considering discipline-specific variations, and teaching the different functions of hedging can contribute to the development of more proficient academic writers who can effectively express uncertainty and mitigate potential contradictions in their writing.

6. Conclusion

The present study has examined the usage of hedging devices in research articles in the soft sciences written by Indonesian professional authors. The overall distribution of hedging devices in the corpus indicated a relatively low frequency. Modal auxiliary verbs emerged as the most frequently used form of hedging devices, with can being the predominant hedging item across all types of hedges. However, it differs from the usage patterns of native English speakers, who tend to favour may or would as hedging devices to introduce uncertainty and mitigate potential contradictions. This difference may reflect cultural and linguistic variations in hedging strategies, with Indonesian writers focusing more on expressing possibility or capability. The study also highlighted the different functions of hedging devices, including accuracy-based, writer-based and reader-based hedges. These findings can inform English language teaching and academic writing instruction by illustrating the diverse functions of hedging and providing students with explicit instruction on their appropriate use in various contexts.

While this study provides valuable insights into the usage of hedging devices in Indonesian research articles in the soft sciences, there are several limitations that should be acknowledged. First, the study focuses only on soft sciences research articles authored by Indonesian writers and published between 2019 and 2022. It is important to acknowledge that the usage of hedges may vary across different disciplines, times and cultural contexts. Therefore, the findings may not be generalisable to other fields or contexts. Future research may investigate the usage of hedging devices by Indonesian writers in different genres of academic writing, such as literature reviews, empirical studies, or case studies. Analysing the variation in hedging practices across genres can provide a nuanced understanding of how hedging functions in different types of scholarly communication. Second, while the study mentions that the usage of hedges by Indonesian authors is relatively low compared to writers of the internationally renowned journals, as shown by previous research, it does not provide a detailed comparative analysis. It would be beneficial for future studies to use a comparative corpus consisting of articles written by L1 English, allowing for a more thorough and explicit comparison. Also, future studies can focus on examining the deployment of hedges in specific sections of research articles, such as the introduction, literature review, methodology, results, and discussion sections. This detailed analysis can provide insights into how hedging devices are used to introduce background information, present research findings, or interpret results within different sections of academic articles.

Correction

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Disclosure statement

No potential conflict of interest was reported by the author(s).