Expression of Epistemic Stance in EFL Chinese University Students ’ Writing

This paper reported findings on a contrastive analysis of epistemic expressions in argumentative essays between NS and NNS Chinese L2 writers. Based on an examination of a NS corpus and a NNS learner corpus across four proficiency levels, the study shows there is great similarity in the total number of epistemic devices used per thousand words between the NS and NNS university students’ written texts. However, the NNS were found to use a restricted range of epistemic items to express epistemic modality and their written texts were marked by strong, unwarranted assertions compared with those of NS students. The study suggested there seems to be an improvement in selecting an appropriate level of commitment to the propositions being expressed with the increase in the students’ proficiency.


Definition of Epistemic Modality
A great deal of work has been done trying to define the term 'epistemic modality'.Derived from Greek word episteme, epistemic modality is concerned with "knowledge and belief" (Lyons 1977;Kärkkäinen, 2003).Coates (1995) argues epistemic modality is "concerned with the speaker's assumptions, or assessment of possibilities, and, in most cases, it indicates the speaker's confidence or lack of confidence in the truth of the proposition expressed" (p.55).Palmer (1986) refers to epistemic modality as an "indication by the speaker of his (lack of ) commitment to the truth of the proposition expressed" and "as the degree of commitment by the speaker to what he says" (p.51).Some researchers adopt a broader definition.For example, Holmes (1982) refers to epistemic meaning as degree of certainty.Biber et al. (1999) argues epistemic stance markers are used "to present speaker comments on the status of information in a proposition" (p.972).

Importance of Study on Epistemic Modality
The ability to express epistemic stance appropriately in English is considered as crucial for successful interaction between the writer and the reader.Epistemic stance can be used by writers to realize diverse social and pragmatic functions (Holmes, 1982;Coates, 1990;Nikula, 1996).However, owing to the great complexity of the epistemic devices available, even NS writers may face great challenge when choosing an appropriate epistemic device to use.Hyland and Milton (1997) identified several reasons why students' manipulation of epistemic devices are so problematic, and these reasons include: 1) the polypragmatic nature of modal expressions; 2) epistemic modality's double functions as both conveying the writer's commitment to the statements and negotiating relationship with readers; 3) epistemic meanings can be expressed in many ways.For English language learners and novice writers in particular, the difficulty of conveying a proper degree of confidence in the truth of the statements they make can be even greater.It is thus very important for language teachers to understand the problems students have with epistemic stance use in their writing, and address those problems accordingly in order to help them improve their academic writing.Unfortunately it is only recently that the study on Chinese EFL students' epistemic stance use began to receive attention in higher education in China, it thus remains largely under-researched what Chinese university students' expression of stance is like compared with that of native students' in English writing.

Review of Literature
Many researchers have carried out comparative studies on the expression of epistemic stance between NS and NNS writers.Hyland and Milton (1997) examined the argumentative texts of NS and NNS Hong Kong high school leavers, and revealed that L2 writers relied on a limited range of epistemic items, made stronger assertions and had greater problems in qualifying the statements in their writing.Milton and Hyland (1999) extended their study to compare the use of lexical phrases between the NS and NNS Chinese students, and similar findings were reported that the NNS students employed a limited number of multi-word hedging and boosting expressions.It was also suggested that the students appear to approximate native-like usage in tentative expressions as their proficiency improves.Chen (2010) also explored the difference in the use of nine epistemic words between NS and NNS Chinese students and found the NS writers used significantly more epistemic devices than the NNS writers, but it appears that the NNS students experience a progress in intercultural pragmatic competence with their increase in language proficiency.These researches all shed some light on how Chinese L2 writers express epistemic modality, but Hyland & Milton's studies deal with high school students, and Chen's research is based on only 9 epistemic items.What the whole picture of the epistemic expression for Chinese university students' writing is like is still under researched.Not only Chinese L2 writers, many researchers have revealed in their studies that native Arabic L2 writers, French and Dutch L2 writers had similar problems in expressing modality with appropriate level of commitment (Scarcella & Brunak, 1981;Dudley-Evans, 1992).Some researchers have also studied epistemic modality in spoken languages.Kärkkäinen (1992) reported findings compatible with those of Hyland and Milton (1997) on written texts.In the study on epistemic expression in L1 and L2 of the same speakers, Letica (2009) found speakers used epistemic devices less frequently in L2 than in L1 and they relied on a limited range of devices in both languages.

Research Question
In order to explore the use of epistemic stance in Chinese EFL university students' writing, the current study aims to address two research questions: 1. What are the characteristics of the expression of epistemic stance in Chinese EFL university students' writing compared with those of native students?2. Is there any difference in students' expression of epistemic stance across different proficiency levels?

Corpora
This study adopted a contrastive learner corpus approach.The two corpora selected are a native English speaker corpus, Louvain Corpus of Native English Essays (LOCNESS) and a learner corpus, the Chinese learner corpus (CLEC).Table 1 shows the total number of word counts in the NS and NNS sub-corpora.).Since it is generally acknowledged English majors have stronger English abilities than non-English majors, we may assume there is an increase in English proficiency across the five levels from ST2 to ST6.The sub-corpus each consisted of about 210 000 words, similar to the size the NS corpus.As this study only focuses on university students' written texts, the high school students' texts labeled as ST2 were not selected.A breakdown of the exact number of words in each sub-corpus is shown in Table 1.Interestingly enough, there are considerable similarities in the total number of epistemic devices used by NNS and NS students in the two corpora.As is shown in Table 2, both groups employed about 26~27 epistemic expressions per 1000 words.It appears that both the NS and NNS writers were aware of the expression of epistemic modality in writing and could employ those devices when necessary.Similar findings have also been reported by Hyland and Milton (1997) in their research on epistemic use of high school leavers in Hong Kong and Britain.The total number of devices used in the two corpora is shown in Table 2.

Expression of Epistemic Stance in NNS Corpus and NS Corpus
Figure 1 shows the distribution of epistemic items used in different grammatical classes across the five proficiency levels.We can find that modal verbs are the most preferred devices to express epistemic modality for both the NS and NNS students.Adverbs are the second most frequently used epistemic devices.There doesn't seem to be a consistent increase or decline in the number of epistemic devices in each grammatical category across the four NNS proficiency bands.However, if we group ST3 and ST 4 as lower level students, ST5 and ST6 as higher level ones, we may notice a significant decline in the frequency count of modal verbs while in the meantime adjective and noun epistemic items rise accordingly.The findings seem to suggest that as students' language proficiency improves, they might be able to select a wider range of epistemic devices to express epistemic modality rather than relying heavily on a restricted number of modal auxiliaries.Table 3 lists the most frequent epistemic items in the NS and NNS corpus in terms of grammatical classes.Surprisingly, both the NS and NNS students favored quite similar range of epistemic items in each grammatical category.For example, will and should both appear on the top of the list for epistemic modals.Similarly, there is also a lot of overlap in the most frequent epistemic adverbs and lexical verbs between the NS and NNS corpora.Despite the great similarities, a number of mismatches can also be found between the two corpora.Most noticeably in the adverbs category, the top six epistemic adverbs account for a little less than 50% of the total adverbials used to express stance in the NS corpus.By contrast, the six most frequent adverbs in the NNS corpora constitute over 60% to 70% of the total.The only exception is the ST6 sub-corpus which indicated a similar percentage to that of the NS corpus.This confirmed the finding that NNS students seem to employ a restricted range of devices to express epistemic modality evidenced by a heavy reliance on a small number of lexical words (Hyland & Milton, 1997).However, there is clearly a decline in the total percentage of the top six epistemic adverbs across the four proficiency levels.The lower level students reported a total of around 70%, and the number dropped to about 61% for ST5 until the highest level students reached a native-like level of less than 50%.Similar development can also be traced in the total percentage of the top three lexical verbs.One possible explanation for the decline is that as students' proficiency increases, they have a wider range of lexical devices at hand to express modality.Students in the lower bands, on the other hand, much as they wanted to, their language proficiency would very likely put constraints on their choice of epistemic devices, and they may feel more confident in selecting those "lower stake" items from the most frequent item list.
Let's take lexical verb 'think' as a example.'Think' ranks the third in the NS corpus, accounting for 16.6% of the total epistemic devices in use.By sharp contrast, the word 'think' tops the list of most frequent epistemic lexical verbs in both the ST3 and ST4 corpus, with a high percentage of 43.2% and 51.8%.In ST5 and ST6 corpus, the percentage dropped to 27.6% and 26.3%, but still they were about 10% higher than the corresponding percentage in the NS corpus.Similar findings were reported by Wang and Zhang (2007) who found Chinese students overused 'think' and 'I think' in their argumentative essays compared with both NS writers and L2 writers of other languages.It may suggest the NNS students had mastered a very limited number of lexical epistemic devices, and 'think' was one that was most familiar to them.Expert writer's written texts are characterized by an appropriate degree of commitment to the propositions they make, but novice writers tend to make unwarranted assertions without qualifying their statements.In order to examine the NS and NNS students' expression of certainty and doubt in the two corpora, we put all the epistemic items into two broad categories: certainty and doubt.Items expressing full commitment to the truth or the proposition were grouped under 'certainty' category, the rest of the items were grouped under 'doubt'.Examples of certainty and doubt items are illustrated in Table 4.

Expression of Doubt and Certainty in NS and NNS Corpora
Figure 2 illustrates how the NS and NNS students in the two corpora express doubt and certainty.It can be seen from the graph that the NS texts are marked by more doubt than certainty expressions whereas NNS students make much stronger assertions in the written discourse.The discrepancy is the biggest in the lowest band ST3 corpus in which the number of certainty items is almost three times that of doubt items.ST4 students also used twice as many certainty items as doubt items.It appears students' ability to qualify their propositions improves with the progress of language proficiency.NNS writers in higher band ST5 and ST6 corpora used significantly less certainty items than those in the lower bands.It might be that as the students' English proficiency increases, they are more aware of the illusionary force the statements carry and thus pay more attention to choosing an appropriate level of commitment to their propositions.

Conclusion
The ability to appropriately express doubt and certainty in written discourse is vital to successful academic written communication.From the examination of the NS corpus and NNS corpus in our study, we conclude :1) Chinese university students' written texts were characterized by stronger assertions compared with those of NS writers; 2) NNS writers used a more restricted range of devices to express epistemic modality than NS writers; 3) students' ability to make appropriate assertions and qualifications showed progress with their language proficiency.There are several reasons that might explain the disparities between the two corpora.One possible reason could be that NS writers have been taught in their writing class to make appropriate assertions and qualifications, but it is still relatively rare to teach the proper use of stance expressions in English writing class in China.Writing is still mainly focused on such traits as organization, grammatical correctness, syntactic variety, and little attention is paid to appropriate stance expression in English writing.Cultural difference might be another possible reason to account for the difference between the two corpora.Although Chinese culture is known to be less assertive, but in Chinese academic writing it is very common to make strong assertions (Long  & Xu, 2010).In their study on Chinese students' use of stance expressions, Long and Xu (2010) found there is a high correlation in epistemic items expressing certainty between the English written texts and the corresponding Chinese written texts of the same topic by the same students.It is quite likely that Chinese students' overuse of assertive expressions is due to L1 transfer.
For NNS students to develop pragmatic competence in writing, language teachers need to be aware of the importance of expressing doubt and certainty in academic writing in the first place.Although recent years have seen a surge in the studies on stance use of Chinese students, there's still a long way to go before the research findings can be materialized in classroom applications.It is therefore necessary for language teachers and educators to design and develop appropriate materials to train the students at various proficiency levels in the proper use of epistemic devices.At the minimal level, attention should be given to the difference in expressing epistemic modality between Chinese and English written discourse to reduce the language transfer that might come into play in their EFL writing.

Figure 1 .
Figure 1.Distribution of epistemic items in different grammatical classes (per 1000 words)

Figure 2 .
Figure 2. Distribution of epistemic items expressing certainty and doubt across proficiency levels

Table 1 .
Total Number of Word Counts in NS and NNS Corpus The total number of words is about 209 783, with 60 209 words of British A-level argumentative essays and 149 574 words of argumentative essays written by American university students.
LOCNESS corpus was used as a NS corpus to carry out the contrastive corpus analysis with the learner corpus.For the purpose of the current study, only argumentative essays of NS students in the corpus were selected for contrastive study.

Table 2 .
Total epistemic devices used in NNS corpus and NS corpus

Table 3 .
Most frequent epistemic items in major grammatical classes (per 10,000 words)

Table 4 .
Examples of certainty and doubt items