1.1Discourse markers

“Discourse markers” (DMs) is one of the labels used (e.g. by Jucker and Ziv 1998Jucker, Andreas H., and Yael Ziv (eds) 1998 Discourse Markers: Descriptions and Theory. Amsterdam/Philadelphia: John Benjamins. ; Schiffrin 1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. ; Schourup 1999Schourup, Lawrence 1999 “Tutorial Overview: Discourse Markers.” Lingua 107: 227–265. ) for the seemingly empty linguistic items with high frequency in oral discourse, such as well, I mean, like, oh, and y’know.11.Another commonly used label for these linguistic items is “pragmatic markers” (PMs) (e.g. by Andersen 2001Andersen, Gisle 2001 Pragmatic Markers and Sociolinguistic Variation: A Relevance-theoretic Approach to the Language of Adolescents. Amsterdam/ Philadelphia: John Benjamins. ; Brinton 1996Brinton, Laurel J. 1996 Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin/New York: Mouton de Gruyter. ). These two terms reflect the different attitudes of the authors to the question of the uniformity of the class of these expressions (Jucker and Ziv 1998Jucker, Andreas H., and Yael Ziv (eds) 1998 Discourse Markers: Descriptions and Theory. Amsterdam/Philadelphia: John Benjamins. ). A general agreement is that PM focuses on “the absence of apparent semantic meaning and grammatical category” of the expression (Wei 2011aWei, Ming 2011a “Investigating the Oral Proficiency of English Learners in China: A Comparative Study of the Use of Pragmatic Markers.” Journal of Pragmatics 43: 3455–3472. , 3455). As suggested by Schiffrin (1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. , 328), common features of DMs are that: they are “syntactically detachable from a sentence;” “have a range of prosodic contours;” and that they either have to “have no meaning, a vague meaning, or to be reflexive (of the language, of the speaker)”. Although DMs are syntactically optional, they are multifunctional. For example, they can help a speaker to organize, hold turns, and mark boundaries in a discourse (Stenström 1994Stenström, Anna-Brita 1994 An Introduction to Spoken Interaction. London and New York: Longman.). They can also help a speaker express attitudes such as uncertainty or hesitation, and to signal politeness (Aijmer 2002Aijmer, Karin 2002 English Discourse Particles: Evidence from a Corpus. Amsterdam/ Philadelphia: John Benjamins. ).

The use of DMs is relevant to second/foreign language (L2) speakers’ communicative needs and pragmatic competence. Comparative studies on DM use by L2 speakers and native speakers have boomed in recent years (e.g. Aijmer 2011Aijmer, Karin 2011 “ Well I’m not sure I think… The Use of Well by Non-Native Speakers.” International Journal of Corpus Linguistics 16: 231–254. on well; Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. on so; Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. on you know, like, so, well). Overuse, underuse, and misuse of L2 DMs have been reported in the literature even with advanced learners of English (e.g. Aijmer 2004Aijmer, Karin 2004 “Pragmatic Markers in Spoken Interlanguage.” Nordic Journal of English Studies 3: 173–190. ), leading to problems of coherence or appropriateness of their speech (Aijmer 2011Aijmer, Karin 2011 “ Well I’m not sure I think… The Use of Well by Non-Native Speakers.” International Journal of Corpus Linguistics 16: 231–254. ; Moreno 2001Moreno, Ángela Eugenia Iglesias 2001 “Native-Speaker – Non-Native Speaker Interaction: The Use of Discourse Markers.” ELLA 2: 129–142.). It is not surprising, therefore, that DMs have remained an area of concern in L2 development of communicative competence. To shed light on learners’ problems, it would also be essential to uncover factors which affect L2 DM use.

In response to the above, the present study makes an attempt to examine the use of but and so by native and Chinese speakers of English because their pragmatic use, especially by Chinese speakers of English, is under-studied. The study also focuses on two influential factors (native language and oral proficiency) for the linguistic behavior of Chinese speakers of English. Although the sample size of this study is small, with five native English speakers and ten Chinese speakers of English, the results of this preliminary study provide a possible direction of research on L2 development of communicative competence.

3.1Communicative competence and L2 DMs

The term “communicative competence” first defined by Hymes (1972)Hymes, Dell 1972 “On Communicative Competence.” In Sociolinguistics, ed. by John B. Pride, and Janet Holmes, 269–293. Harmondsworth: Penguin. as knowledge of what is possible, feasible, appropriate and actually done has been developed by a number of researchers. The key components of communicative language ability discussed in this section are based on the synthesis of existing models in the literature (Bachman 1990Bachman, Lyle F. 1990 Fundamental Considerations in Language Testing. Oxford: Oxford University Press.; Bachman and Palmer 2010Bachman, Lyle F., and Adrian S. Palmer 2010 Language Assessment in Practice. Oxford: Oxford University Press.; Canale and Swain 1980Canale, Michael, and Merrill Swain 1980 “Theoretical Bases of Communicative Approaches to Second Language Teaching and Testing.” Applied Linguistics 1: 1–47. ; Hedge 2000Hedge, Tricia 2000 Teaching and Learning in the Language Classroom. Oxford: Oxford University Press.). It consists of four components: linguistic competence, pragmatic competence, discourse competence, and strategic competence.

Linguistic competence includes knowledge of pronunciation, vocabulary, morphology, semantics and syntax of the language. Only ideational meaning of DMs is related to linguistic competence but pragmatic use of DMs is not, since PMs are syntactically optional. Without PMs, the language remains grammatically correct.

Pragmatic competence includes two kinds of ability: illocutionary competence, which means knowing how to use language in order to achieve certain communicative goals; and sociolinguistic competence, which is knowledge of the appropriate social conventions of how to use the language in varying situations in a particular culture. Not using DMs in one’s speech will not lead to pragmatic failure since DMs are syntactically optional. However, L2 speakers need to master the pragmatic functions of DMs, not only the grammatical forms, for the negotiation of the relationship between speaker and hearer during a conversation (Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. ). Furthermore, L2 speakers need to know that DMs are widespread in natural speech (Ostman 1982Ostman, Jan Ola 1982 “The Symbiotic Relationship between Pragmatic Particles and Impromptu Speech.” In Impromptu Speech: A Symposium, ed. by Nils Enkvist, 147–177. Abo, Finland: Abo Akademi.) and vary in frequency and type according to formality (e.g. Aijmer 2002Aijmer, Karin 2002 English Discourse Particles: Evidence from a Corpus. Amsterdam/ Philadelphia: John Benjamins. ; Jucker and Smith 1998Jucker, Andreas H., and Sara W. Smith 1998 “ And People Just you Know like ‘Wow’: Discourse Markers as Negotiating Strategies.” In Discourse Markers: Descriptions and Theory, ed. by Andreas H. Jucker, and Yael Ziv, 171–201. Amsterdam/Philadelphia: John Benjamins. ).

Discourse competence, or textual competence, includes knowledge of joining utterances together to form a coherent text or conversation. Use of DMs in this sense relates to L2 speakers’ discourse competence, particularly the ability of language learners to take longer turns, and open and close conversation. For example, language learners need to learn the DMs which signal continuation of discourse such as and, and succession of discourse such as then.

Strategic competence refers to a speaker’s ability to exploit communication strategies in an authentic communicative situation. One of the communication strategies is using hesitation devices to maintain the conversation. Here, use of DMs relates to L2 speakers’ strategic competence as one function of DMs is to serve as a filler or delaying tactic (when searching for appropriate words for a later utterance).

3.2Previous studies on ‘but’ and ‘so’ by L2 learners

Although but and so used by L2 English speakers are less studied than some DMs such as like, well, and you know, several scholars have published findings on their use. Hays (1992)Hays, Paul R. 1992 “Discourse Markers and L2 Acquisition.” Papers in Applied Linguistics-Michigan 7: 24–34. found that Japanese speakers of English used and, but, and so at much higher rates than other markers such as because, well, I mean, oh, and you know. He labeled and, but, and so ideational markers because he claimed that these markers related to text not pragmatics. Fung and Carter (2007)Fung, Loretta, and Ronald Carter 2007 “Discourse Markers and Spoken English: Native and Learner Use in Pedagogic Settings.” Applied Linguistics 28: 410–439.Emerald Group compared the use of English DMs by Hong Kong speakers and native British English speakers, and found that Hong Kong learners displayed a frequent use of referential markers (and, but, because, ok, so, etc.), but a more restricted use of interpersonal markers (yeah, really, say, sort of, I see, you see, well, right, actually, you know, etc.). They suggested that this result reflected that the learners were exposed to traditional grammar-centered pedagogy and hence focused on the semantic meanings of words rather than their pragmatic use in spoken language. Wei (2011b)Wei, Ming 2011b “A Comparative Study of the Oral Proficiency of Chinese Learners of English across Task Functions: A Discourse Marker Perspective.” Foreign Language Annals 44: 674–691. investigated the use of DMs by university learners of English in China at different proficiency levels (intermediate and advanced) across four task functions. She found that the top four most frequently used DMs were and, but, also and so in the same order by both groups.

Two further studies comprehensively examined the use of so by English learners from Europe. Müller (2005)Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. found that native English speakers used the textual functions of so significantly more often than German speakers of English; however, the interactional functions and the ideational functions of so did not show any significant difference. Buysse’s (2012)Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. study showed that Dutch speakers of English used so at a much higher rate than native speakers of English, and that they used interpersonal and textual so at higher rates than native speakers on the whole.

The current study’s first research goal is to examine the ideational/pragmatic use of but and so by Chinese speakers of English and native English speakers for three reasons. The literature review shows that but and so occur most frequently in the speech of both native and non-native English speakers (e.g. Fung and Carter 2007Fung, Loretta, and Ronald Carter 2007 “Discourse Markers and Spoken English: Native and Learner Use in Pedagogic Settings.” Applied Linguistics 28: 410–439.Emerald Group ; Hays 1992Hays, Paul R. 1992 “Discourse Markers and L2 Acquisition.” Papers in Applied Linguistics-Michigan 7: 24–34.). In addition, use of but by L2 speakers is under-studied in the literature. While the metafunctions of so by L2 learners have been examined by studies such as Buysse (2012)Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. , and Müller (2005)Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. , none of the previous studies have examined its use by Chinese speakers of English. If participants’ first language (L1) is different, the effect of L1 transfer on L2 DMs would be different, and their use of L2 DMs would also be different.

The current study also sets out to explore how L2 speakers’ L1 (Mandarin Chinese) and overall oral proficiency may affect their ideational/pragmatic use of but and so, to better understand the linguistic behavior of L2 speakers. The need of research about L1 influence on the use of L2 DMs has been noticed by several researchers in the field (e.g. Baumgarten and House 2010Baumgarten, Nicole, and Juliane House 2010 “ I think and I don’t know in English as Lingua Franca and Native English Discourse.” Journal of Pragmatics 42: 1184–1200. ; Liao 2009Liao, Silvie 2009 “Variation in the Use of Discourse Markers by Chinese Teaching Assistants in the US.” Journal of Pragmatics 41: 1313–1328. ; Müller 2004Müller, Simore 2004 “ ‘Well you Know that Type of Person’: Functions of Well in the Speech of American and German Students.” Journal of Pragmatics 36: 1157–1182. ; Wei 2011b)Wei, Ming 2011b “A Comparative Study of the Oral Proficiency of Chinese Learners of English across Task Functions: A Discourse Marker Perspective.” Foreign Language Annals 44: 674–691. , yet it remains an under-studied area in DM studies. Additionally, according to Wei’s (2011b)Wei, Ming 2011b “A Comparative Study of the Oral Proficiency of Chinese Learners of English across Task Functions: A Discourse Marker Perspective.” Foreign Language Annals 44: 674–691. study, L2 speakers’ overall oral proficiency level relates to the way L2 DMs are used. However, that study did not examine the relationship of oral proficiency and ideational/pragmatic use of but and so because the goal was to investigate the relationship between L2 speakers’ oral proficiency level and L2 DM use across different contextual situations.

The current study therefore addresses the following two research questions:

1. What are the similarities and differences in the ideational and pragmatic use of but and so by native English and Chinese speakers of English?

2. In what ways do L1 and the overall oral proficiency of Chinese speakers of English influence their ideational and pragmatic use of but and so?

4.1Participants

The data of this study are from a larger project which described the use of English DMs by L1 Chinese speakers of English (Liu 2009Liu, Binmei 2009 Discourse Marker Use by L1 Chinese EFL Speakers. The University of Florida Ph.D. Thesis.). The ten L2 English participants involved in that study were graduate students (five male and five female) at an American university, originally from mainland China, and aged between twenty-two and thirty-seven years old (mean = 28.2). The participants have been assigned Chinese pseudonyms (Dong, Feng, Bing, Lian, Xia, Qiu, Peng, Jun, Fang, and Juan) to maintain confidentiality. Each one had received formal English education in China for more than eight years, and had been studying various programs including education, tourism, computer science, linguistics, and biochemical engineering in the U.S. for between three weeks and four years. Their English proficiency was considered advanced because each of the participants scored over 600 in TOEFL (Test of English as a Foreign Language) when the data were collected.

Each participant was asked to take the TSE (Test of Spoken English) at the American university, and they were divided into two groups according to their scores: a higher oral proficiency group, and a lower oral proficiency group. The higher oral proficiency group had a minimum 45 in the TSE (they are Dong, Feng, Lian, Xia, Qiu and Peng) whereas the lower oral proficiency group obtained a 40 in the TSE (they are Bing, Jun, Fang, Juan). All members of the higher proficiency group except Peng had studied in the U.S. or Canada for over a year when the data were collected, while all those in the lower oral proficiency group, except Bing, had studied in the U.S. for just three weeks.

Five American English native speakers participating in the study (two male and three female) were graduate students at the same university. They studied linguistics, biomedicine and English, and their ages ranged from twenty-four to thirty-five (mean = 29.2). The American participants have been assigned American pseudonyms (John, Davy, Sue, Ann, and Mary).

4.2Data collection

Individual sociolinguistic interviews were conducted to elicit DMs. English interviews were conducted by an American (pseudonym Sherry) who was a graduate student at the same American university. Each participant, native or non-native English speaker, was interviewed for about fifteen minutes. Topics for the interviews were personal in order to elicit an oral narrative register. Interviews with L1 English and L2 English speakers followed the same interview questions (such as weekends, hobbies, travels, favorite sports, favorite teachers, favorite movies, favorite TV programs, and happy experience). Two or three weeks after the English interviews were completed, the ten L1 Chinese speakers were interviewed by the researcher in Mandarin Chinese. The researcher was a graduate student from the same American university, and she also came from mainland China. Each participant was interviewed for about fifteen minutes. The Chinese interviews used the same questions as the English ones.

The data included fifteen interviews in English and ten interviews in Chinese. However, the total number of words in the transcripts included approximately 31,000 English words and 33,500 Chinese words. Indeed, one of the limitations of many DM studies, including the current study, is small sample size. The transcription conventions in the current article follow Müller (2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. , 281) (see Appendix A). The principles used are simpler than Müller’s in that vocal noises, quality, and code switching were not included because they are not related to the research questions under investigation. The next section will sketch the functions of but, so, and their corresponding Chinese expressions that have been attested in the data.

5.1Categorizing ‘but’

Schiffrin (1987)Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. states that the primary function of but is its ideational function – marking the contrast of ideas, because it is a conjunction. She further suggests that but has a textual function, as a marker of speaker-return – marking claims for the floor and allowing the speaker to return to the point (Schiffrin 1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. ). Bell (1998)Bell, David M. 1998 “Cancellative Discourse Markers: A Core/periphery Approach.” Pragmatics 8: 515–541. calls this a sequential function of but which signals a return to the main topic or point, and Norrick (2001)Norrick, Neal R. 2001 “Discourse Markers in Oral Narrative.” Journal of pragmatics 33: 49–878. applies the speaker-return or shifting function of but in the oral narrative genre.

When it contains an ideational meaning, but is often stressed in the data; however, when serving textual functions, it is never stressed. Analysis of the data in this study reveals but was found to serve one ideational and two textual functions.

The ideational function of but marks the contrast of ideas, as shown in Example (1). When Sue was talking about her favorite professor, but marked the disagreement with the speaker’s own previous discourse. It has an ideational meaning, and it is not semantically bleached, although it is syntactically optional.

The first textual function is the use of but as a marker of speaker-return shown in Example (2). John at first said they got on a train and he was sure the train had a themed name, then he said he didn’t remember the name. However, he started to describe the train in the following utterance where the underlined but loses its contrastive meaning, and its function is to allow the speaker to return to his original point – the train.

The second textual function of but is holding the floor for the speaker.66.Floor-holding is considered a textual function by some scholars (e.g. Brinton 1996Brinton, Laurel J. 1996 Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin/New York: Mouton de Gruyter. ), while an interpersonal function by others (e.g. Aijmer 2002Aijmer, Karin 2002 English Discourse Particles: Evidence from a Corpus. Amsterdam/ Philadelphia: John Benjamins. ; Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. ). This function is illustrated in Example (3), in which Mary was telling the interviewer what places she had been to. At the end, Mary gave a summary (“that’s the extent of my travel”), filled by but. Here but loses its contrastive meaning, and instead, functions as a floor-holder allowing the speaker to have more time to search for what he/she wants to say next. There is always a pause after this use of but.

In summary, but was found to serve one ideational function and two textual functions in the data. When serving the ideational function, but marks the contrast of ideas as shown in Example (1), whereas the textual functions serve as a speaker-return marker and a floor-holder for the speaker, illustrated in Examples (2) and (3) respectively.

5.2Categorizing danshi ‘but’

The translation equivalent of but has three forms (‘dan/danshi/buguo’) in the Chinese interview data. They are interchangeable conjunctions, and the most common form in the data is ‘danshi’. Fang (2000)Fang, Mei 2000 “Ziran kouyu zhong ruohua lianci de huayu biaoji gongneng [Reduced conjunctions as discourse markers]”. Zhongguo Yuwen [Chinese language] 5: 459–470. suggests that danshi loses its contrastive meaning when used as a DM and it serves two functions: shifting a topic and holding the floor.

In my data, two functions of danshi were found to be used: the ideational function – marking contrast; and a textual function – marking speaker-return. The floor-holding function of danshi was not found to be used by the L1 Chinese speakers. Example (4) illustrates the two functions of danshi when Lian was talking about her travels in China. In the first Question-Answer turn, Lian said she wanted to travel; however, she didn’t have enough time. The first underlined danshi marks contrastive ideas serving the ideational function. In the second turn, since the two questions are closely related, Lian used danshi to signal a return to her original point, and this time danshi loses its contrastive meaning.

(4)77.For abbreviations of the Mandarin Chinese gloss when there is no lexical English equivalent, please see Appendix B (Li and Thompson 1981Li, Charles. N., and Sandra A. Thompson 1981 Mandarin Chinese: A Functional Reference Grammar. Berkeley/Los Angeles/London: University of California Press., xxiii).

Interviewer:

zai	guo	nei	de	shihou	qu	guo	shenme	difang	wan	ma?
when	country	in	nom	time	go	past	what	place	play	prt

‘When you were in China, which place did you travel to?’

Lian:

(…)

qishi

wo

hen

xiang

wan,

danshi

mei

you

shijian.

Ranhou

Beidaihe

qu

guo,

Tianjin

qu

guo.

actually

I

very

want

play

but

no

have

time

and

Beidai River

go

exp

Tianjin

go

exp

‘Actually I really wanted to travel, but I didn’t have time. And I have been to the Beidai River, been to Tianjin.’

Interviewer:

you	shenme	tebie	youyisi	de	defang?
have	what	special	interesting	nom	place

‘What is the most interesting place you’ve been to?’

Lian:

danshi

zai

yinwei

wo

zai

chu

guo

zhi

qian

qu

de

nage.

Jiuzhaigou

he

nage..

Leshandafo.

Ranhou

wo

ganjue

ting

haode.

but

at

because

I

at

leave

country

nom

before

go

nom

that

Jiuzhai Valley

and

that

Leshan Giant Buddha

and

I

feel

very

good

‘But because before I went abroad, I had been to that.. the Jiuzhai Valley, and that.. the Leshan Giant Buddha. And I feel that is very interesting.’

The Chinese DM danshi was therefore found to serve an ideational meaning and a textual function in this study. When having the ideational meaning, danshi marks contrast; and when serving the textual function, it marks speaker-return, with both uses illustrated in Example (4). Furthermore, the data suggest that these functions are similar to the uses of its translation equivalent but except that but was found to have a floor-holding function while danshi was not.

5.3Categorizing ‘so’

Schiffrin (1987)Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. suggests that the ideational function of so conveys a meaning of result. She also proposes that “so has a pragmatic use in participation structures – as a marker of potential transition” (1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. , 227). Müller (2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. , 68) investigated so on a “textual level” (marking result or consequence, main idea unit marker, summarizing/rewording/giving an example, sequential use, and boundary marker) and an “interactive level” (speech act marker-opinion, question or request, marking implied result, and marker of a transition relevance place). Lam (2010) analyzed use of so by Hong Kong speakers of English in different text types. The discourse functions of so in his study were analyzed in textual functions (framing, linking, and consequential), the interpersonal function (responsive), and interactional functions (processing, and turn managing). Buysse (2012)Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. classified so according to three metafunctions: ideational (indicating a result), interpersonal (drawing a conclusion, prompting, and holding the floor), and textual (introducing a summary/a section of the discourse/a new sequence/elaboration, indicating a shift, and marking self-correction).

In this study, the DM so serves one ideational function, four textual functions, and one interpersonal function. The ideational function of so indicates a result, as in Example (5). Davy said he was Jewish and he went to Israel when he was thirteen because it was traditional. The utterance after so (“I went when I was thirteen”) is the result of the previous utterance (“it’s sort of sort of traditional at that time to go to Israel”). So indicates a result/consequence here, and it can be substituted by “therefore.” This use of so is syntactically optional yet semantically unbleached.

The first textual function of so in the data is to introduce a summary, or to summarize what the speaker has said before. In Example (6), Davy told the interviewer why he chose UF as the graduate school. The first reason was that he wanted to study “Discourse”, and the second reason was the issue of financial aid. In the last sentence of his turn, he summarized his descriptions by saying “so those two things sort of were were reasons why I came here.”

In this function of so, the utterance following so expresses the same propositional idea as the previous discourse (Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. ). The speaker does not posit a new claim, but restates the main arguments in more general terms (Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. ). So in this function is more semantically bleached than the resultative so.

The second textual function of so in the data is to introduce a sequence. It is used to introduce the next event in a series of events, or the next part of the story in a narrative (Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. ) or a new step in an explanation (Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. ). This function is illustrated in Example (7) in which Mary was introducing herself to the interviewer.

Mary said she was born in Canada, and moved to the U.S. when she was three; then her family moved to Florida when she was eleven. The underlined so here was used by the speaker to introduce the next event in her narrative. It is also more semantically bleached than the resultative so.

The third textual function of so indicates a topic shift. It is used when the speaker comes back after a digression or an explanation to a topic or an opinion mentioned before (Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. ). Mary in (8) first said her favorite professor was her mentor, and she started to tell the interviewer that a lot of her courses were multi-subject courses and different professors tried different sub-topics. Then she uses so to return to talk about her favorite professor. In this case, so signals that the speaker shifts back to a higher unit of discourse and used in this way, “either after a brief interruption by or an exchange with the interviewer or after a turn-internal digression” (Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. , 1772).

The fourth textual function of so is floor-holding, and this use of so is always followed by a pause. Lam (2010Lam, Phoenix W. Y. 2010 “The Effect of Text Type on the Use of So as a Discourse Particle.” Discourse Studies 11: 353–372. , 364) suggests that in this function, so is used ‘‘as a delay strategy and signals that the speaker is undergoing some processing problem and requires extra time”. This function is illustrated in Example (9) below.

In this example, the underlined so is followed by a pause, but it does not introduce the following segment. The speaker holds on to the floor with so because he needs to have extra time to start a new segment.

The interpersonal function of so in the data is used to yield the floor,88.Relinquishing the floor is considered a textual function by Brinton (1996)Brinton, Laurel J. 1996 Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin/New York: Mouton de Gruyter. . This use of so is considered an interpersonal one by Buysse (2012)Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. and Müller (2005)Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. . as in Example (10).

This use of so occurs at the end of a turn, followed by no other words, and always carries a falling or level tone (Buysse 2012Buysse, Lieven 2012 “ So as a Multifunctional Discourse Marker in Native and Learner Speech.” Journal of Pragmatics 44: 1764–1782. ). The use of so in this situation is “elliptical” because it “does convey a ‘result’ meaning even if no result follows” (Schiffrin 1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. , 223). This function “gives an additional cue for the hearer that s/he may take over the floor” (Müller 2005Müller, Simore 2005 Discourse Markers in Native and Non-native English Discourse. Amsterdam/Philadelphia: John Benjamins. , 85).

In addition to its resultative ideational meaning explained in Example (5), so has multiple pragmatic functions which include four textual functions and one interpersonal function in the data. Illustrated by Examples (6) to (9), the textual functions are listed as introducing a summary, introducing a sequence, indicating a topic shift, and holding the floor for the speaker. The interpersonal function of so is used at the end of a turn, followed by no other words to indicate that the speaker wants to yield the floor to the hearer, as in Example (10).

5.4Categorizing suoyi ‘so’

The translation equivalent of so has two interchangeable forms in the Chinese data (‘suoyi/suoyishuo’), with ‘suoyi’ the more common form in the data. Suoyi is a conjunction conveying a result meaning in Chinese. Fang (2000)Fang, Mei 2000 “Ziran kouyu zhong ruohua lianci de huayu biaoji gongneng [Reduced conjunctions as discourse markers]”. Zhongguo Yuwen [Chinese language] 5: 459–470. claims that when becoming semantically bleached, suoyi serves three pragmatic functions: indicating a shift back to the original topic, closing a turn, and a strategy for continuing his/her own turn.

In the current study, suoyi was found to serve the following sub-functions: One ideational function – indicating a result; two textual functions – topic shifting and floor-holding; and one interpersonal function – yielding the floor. In Example (11), Juan said the reason she decided to come to the U.S. was that she felt her job in China was boring. The segment (“I felt my work was not interesting”) before suoyi is the reason for the segment after suoyi (“came here”). Suoyi indicates a result/consequence while retaining its semantic meaning.

The first textual function of suoyi is indicating a shift back to the original topic which is similar to the topic-shifting function of so. Suoyi is used when the speaker comes back after a digression or an explanation to a topic or an opinion mentioned before, as in Example (12). Bing first said he felt that Americans lived a more real life than Chinese. He then continued talking about how Chinese focused on finding a job, buying a house, etc. and then he made the conclusion that Chinese only focused on the result instead of enjoying the process. After using suoyi, he returned to the original topic, talking about Americans, saying that they knew how to enjoy themselves.

The second textual function of suoyi is holding the floor. Similar to floor-holder use of so, this function of suoyi is usually followed by a pause in the data. In Example (13), Xia talked about why she was busy teaching during the summer. At first, she told the interviewer that she was very tired every day because she had to prepare for two hours before each class. Then she continued her topic, filled by suoyi, and added more reasons why teaching made her so busy, e.g., marking homework. Suoyi here functions as a delay strategy for the speaker to search for her next utterance.

The interpersonal function of suoyi is yielding the floor, and this use is similar to the interpersonal function of its corresponding English expression so. For example, in (14), Qiu talked about her experience of choosing a major in her university in China. In the first turn she told the interviewer that she didn’t choose Japanese as her major, and in the second turn she at first explained that the leaders of her school persuaded her to Choose Japanese and she then accepted this arrangement as her fate, while afterwards she felt terrific about the choice. Suoyi was used at the end of her second turn, followed by no other words, indicating that Qiu was ready to give the floor to the hearer.

The Chinese DM suoyi was found to have: a resultative ideational meaning; two textual functions – topic shifting and floor-holding; and one interpersonal function – yielding the floor, as shown in Examples (11–14). These uses are similar to those of its translation equivalent so in the data; however, so was also found to serve two more textual functions: introducing a summary; and introducing a sequence.

5.5An overview of qualitative results

Tables 1 and 2 provide an overview of all DM functions identified for but, so, and their corresponding Chinese expressions dan/danshi/buguo and suoyi/suoyishuo in the data, presented according to the type of discourse relation they mark.

Table 1 shows that but was found to serve one ideational function and two textual functions in the data. Its Chinese counterpart was found to serve an ideational meaning and a textual function. The functions of the Chinese counterpart are similar to the uses of but, except that but was found to have a floor-holding function while dan/danshi/buguo was not.

Table 2 indicates that so was found to have one ideational function, four textual functions and one interpersonal function. Its corresponding Chinese expression was found to serve one ideational function, two textual functions and one interpersonal function. The uses of the Chinese counterpart are similar to those of so in the data; however, so was found to serve two more textual functions, i.e. introducing a summary and introducing a sequence. The next section will present the quantitative data by comparing the use of DMs between the L1 and L2 English speakers, and then address how L1 and oral proficiency affect L2 DM use.

6.1Statistical findings by English and Chinese speakers

The coding of the sub-functions of but and so in each participant’s English speech was based on the categorizations in the previous section. The frequency of each function of but and so per person was calculated according to the total tokens per 1,000 words (see Tables 3 and 4).

Table 3 shows that both the native and non-native English speakers used the ideational function of but at the highest rates among the sub-functions. The native speakers used the speaker-return function at a similar rate as the ideational function of but (3.99 vs. 3.33 tokens per thousand words per person), while the use of but as a floor holder was the lowest rate (1.36 tokens per thousand words per person) of all the functions. The non-native English speakers used the ideational function at a much higher rate than the speaker-return function of but (8.15 vs. 1.79 tokens per thousand words per person) and none of them used the floor-holding function of but.

Table 4 shows that both groups used the ideational function of so at much higher rates than other functions, and they both used the sequencing function at the second highest rates of all the sub-functions.

Coding of the sub-functions of their Chinese corresponding expressions in each L1 Chinese speaker’s Chinese data was also based on the categorizations in the previous section. The frequency of each function of dan/danshi/buguo ‘but’ and suoyi/suoyishuo ‘so’ per person was calculated according to the total tokens per 1,000 words (see Tables 5 and 6).

Independent-Samples T tests were run in order to see if there is a significant difference in the uses of but and so by the native English speakers and the Chinese speakers of English; the results are presented in Table 7.

The results in Table 7 show that the Chinese speakers of English used the ideational function of but more often than the native English speakers at a significant level (p < .05). However, the native speakers used the two textual functions of but significantly more frequently than the non-native English speakers. The Chinese speakers of English used but a total number of times at a slightly higher rate than the native speakers (9.94 vs. 8.68 tokens per thousand words per person), but not at a significant level. We can also see that the native English speakers used so at a higher rate than the non-native speakers (13.53 vs. 9.43 tokens per thousand words per person), but not at a significant level. As for the sub-functions, the native speakers used the ideational function of so at a slightly lower rate than the non-native speakers (5.54 vs. 6.61 tokens per thousand words per person). They used the sequencing function (3.61 vs. 0.84 tokens per thousand words per person) and the shifting function (0.48 vs. 0.23 tokens per thousand words per person) at higher rates than the Chinese speakers of English, but not at a significant level. They used the summarizing function more frequently than the non-native English speakers (1.72 vs. 0.51 tokens per thousand words per person), and the difference between them neared significance (p = .059). Both the native and non-native speakers used the interpersonal function ‘floor-yielding’ at similarly low rates (0.60 vs. 0.64 tokens per thousand words per person). Finally, the native speakers used the floor-holding function (1.58 vs. 0.60 tokens per thousand words per person) more frequently than the non-native speakers at a significant level (p < .05).

Independent-Samples T tests were also run to see if there is a significant difference in but and so between the higher oral proficiency group and the lower oral proficiency group of the Chinese speakers of English. Table 8 presents the results.

Table 8 shows that the higher oral proficiency group used the textual and interpersonal functions of but and so at higher rates than the lower oral proficiency group, and the lower oral proficiency group used the ideational but at a much higher rate than the higher oral proficiency group. Furthermore, the higher oral proficiency group used the total number of so at an obviously higher rate than the lower oral proficiency group.

6.2Discussion

Hays (1992)Hays, Paul R. 1992 “Discourse Markers and L2 Acquisition.” Papers in Applied Linguistics-Michigan 7: 24–34. found that but and so were the second and third most frequently used DMs by Japanese speakers of English. Fung and Carter (2007)Fung, Loretta, and Ronald Carter 2007 “Discourse Markers and Spoken English: Native and Learner Use in Pedagogic Settings.” Applied Linguistics 28: 410–439.Emerald Group found that Hong Kong speakers of English used but more frequently than native speakers of English. Wei (2011b)Wei, Ming 2011b “A Comparative Study of the Oral Proficiency of Chinese Learners of English across Task Functions: A Discourse Marker Perspective.” Foreign Language Annals 44: 674–691. also found that but and so were among the most frequently used DMs by both intermediate and advanced Chinese learners of English. However, none of these studies compared but/so in ideational and pragmatic functions. The results of this study suggest that Chinese speakers of English only used the ideational function of but and so at a higher rate than native English speakers, but not the pragmatic functions which is inconsistent with the findings of Fung and Carter (2007)Fung, Loretta, and Ronald Carter 2007 “Discourse Markers and Spoken English: Native and Learner Use in Pedagogic Settings.” Applied Linguistics 28: 410–439.Emerald Group . The Chinese speakers of English underused almost all of the pragmatic functions of the two DMs (except the floor-yielding function of so) compared with the native English speakers, indicating that the non-native English speakers had not acquired the pragmatic functions of but and so to the same degree as the native speakers even though they were advanced learners of English. Among the non-native English speakers, there is still a gap in discourse competence and pragmatic competence in the use of but and so, especially in the three sub-functions of the two DMs.

None of the Chinese speakers of English were found to use the floor-holding use of but, while four out of the five native English speakers used it. Furthermore, the sequencing function of so is the second most frequently used function of so, and four out of the five native English speakers were found to use it. However, only three out of the ten Chinese speakers of English were found to use this function. In addition, the summarizing function of so is the third most frequently used function of so by the native English speakers, with four out of the five native English speakers found to use it. By contrast, only three out of the ten Chinese speakers of English were found to use this function.

With regard to the research question of influence of L1, there are three pieces of evidence showing the influence of L1 on the use of but and so. Tables 5 and 6 show that the L1 Chinese speakers used the ideational functions of danshi ‘but’ and suoyi ‘so’ at higher rates than their pragmatic functions. This result is consistent with the use pattern of their corresponding English DMs – but and so. The frequent use of the ideational but/so could be because the Chinese speakers of English transferred the ideational functions of danshi ‘but’ and suoyi ‘so’ to the ideational uses of but and so. Especially in the use of the ideational but, the Chinese speakers of English used it significantly more frequently than the native English speakers. Specifically, Bing used both the ideational but and the ideational danshi ‘but’ at the highest rates (see Tables 3 and 5) which is perhaps due to L1 transfer. Furthermore, none of the L1 Chinese speakers in the study used the floor-holding function of but or danshi ‘but’ (see Tables 3 and 5). The non-native English speakers’ lack of use of floor-holding but might be due to their lack of use in their L1. Additionally, the sequencing function and summarizing function of so were the second and third most frequently used functions by the native English speakers respectively. However, not many Chinese speakers of English were able to use these two functions in the data, and Table 6 shows that none of the Chinese speakers used suoyi ‘so’ for sequencing or summarizing either. Therefore, the lack of use of sequencing and summarizing suoyi ‘so’ in Chinese might lead to their low tokens of sequencing and summarizing so in English.

To sum up, the results of this study suggest that the ideational functions of but/so are more easily transferred from L1, perhaps because their corresponding Chinese equivalents are translatable. On the other hand, the pragmatic functions of but/so/danshi ‘but’/suoyi ‘so’ are semantically bleached; therefore, no translation equivalents are available for Chinese speakers of English even when transfer is needed.

In response to the research question of influence of oral proficiency, the DM use patterns between the two oral proficiency groups are quite similar to the patterns between the native and non-native English speakers. Moreover, the finding that the higher oral proficiency group had a more frequent use of the textual functions of but and so coincides with the nature of DMs in that one of their functions is to serve as a hesitation device in colloquial speech (Brinton 1996Brinton, Laurel J. 1996 Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin/New York: Mouton de Gruyter. ), thus they can be indicators of L2 fluency (He and Xu 2003He, Anping, and Manfei Xu 2003 “Zhongguo daxuesheng yingyu kouyu small words de yanjiu [Small words in Chinese EFL learners’ spoken English].” Waiyu Jiaoxue Yu Yanjiu [Foreign language teaching and research] 35: 446–452.; Tsai and Chu 2015Tsai, Pei-Shu, and Wo-Hsin Chu 2015 “The Use of Discourse Markers among Mandarin Chinese Teachers, and Chinese as a Second Language and Chinese as a Foreign Language Learners.” Applied Linguistics: amv057. ). When L2 speakers use DMs for hesitation or for linking words and phrases, speech sounds faster and more fluent than when they pause or use er, and erm too often. Meanwhile, the result that the higher oral proficiency group in the present study used the interpersonal function of so at a higher rate indicates that more proficient L2 speakers may be more capable of managing the social relationship with the addressee, which is consistent with the findings in Wei (2011a)Wei, Ming 2011a “Investigating the Oral Proficiency of English Learners in China: A Comparative Study of the Use of Pragmatic Markers.” Journal of Pragmatics 43: 3455–3472. .