A Comparative Study on the Use of the Discourse Marker “Well” by Chinese Learners of English and Native English Speakers

Discourse markers such as “I mean”, “you know”, “well”, etc. play a vital role in daily communication. Among those markers, “well” is found to be the most frequently used one in conversation. Though lots of research have been conducted on such theoretical plains as their definitions, functions, categorizations, few has been designed to investigate how L2 learners use these markers in conversation and in which way(s) they differ from native speakers. Taking these into consideration, the present paper is designed to reveal how Chinese learners of English use the discourse marker “well” and how the pragmatic functions of this marker are preferred in conversation. Results show that (1) Chinese learners of English significantly underuse the discourse marker “well” in conversation; and (2) in terms of its pragmatic functions, Chinese learners of English only prefer to use its delay marker function and initiation marker function.


Introduction
In spoken interaction, speakers may frequently employ sets of linguistic signals as strategies for implementing successful communication. Discourse markers, "sequentially dependent elements which bracket units of talk" (Schiffrin, 1987: 31), are among such linguistic signals.
With the shift of focus from grammatical competence to communicative competence in language teaching, more and more efforts are directed to the study of such small words as discourse markers. Lots of studies (Levinson, 1983;Schiffrin, 1987;Redeker, 1991;Fraser, 1999;Sperber & Wilson, 1995;Blakemore, 2002) have examined the general attributes of discourse markers; some (Quirk et al., 1972;Jucker, 1993;Ran, 2003) adopt a theoretical account of a certain marker like "well"; and still others (Anderson et al., 1999;Trillo, 2002;Chen, 2002;He & Xu, 2003) give an empirical analysis either of discourse markers in general or of a particular one.
However, these studies have not examined the functional distribution frequencies of the discourse marker "well" employed by native English speakers and how Chinese EFL learners acquire the different usages of the marker "well". In addition, previous studies on discourse markers are mostly qualitative in nature. Admittedly, qualitative studies are "extremely useful when the studies are exploratory in nature" (Wen, 2001), but the findings that have been yielded from this kind of studies are more "subjective and based upon the individual's own internalized and cognitive perception of language" (McEnery & Wilson, 1996: 87). By contrast, quantitative studies can provide more objective statements on actual language use. As a new trend in language research, more and more quantitative studies are based on corpus data because corpus-based quantitative research can result in greater reliability and accuracy for the reason that a corpus is supposed to be an adequate representation of naturally occurring discourses.
Therefore, this study, by adopting a corpus-based approach, is implemented to find out how Chinese learners of English differ from native English speakers in the use of the discourse marker "well" and its pragmatic functions in conversation.

Reconstruction of the Pragmatic Functions of the Discourse Marker "Well"
Based on the research findings of Ran (2003), Jucker (1993), and the corpus data (for details, see Section 3.2) being involved, the present paper proposed the following five major pragmatic functions of the discourse marker "well" in conversation:

As a Delay Marker
In conversation the speaker, from time to time, is not immediately ready to give responses to another utterance, or is reluctant to tell the truth, or needs some more time to organize his/her talk. In these situations and alike, a short silence might occur. However, more frequently, the speaker may use some linguistic signals to hold the floor and to make the seemingly incoherent units of utterances into a coherent one; "well" is among such linguistic signals. If "well" is used in this context, it is said to function as a delay marker, as in Example (1).
(1) A: Oh, come, Mathilde, surely you can tell an old friend. B: Well, … well, it was all because of the necklace.
In the above example, Speaker B apparently needs more time to think about the answer. Consequently, he/she repeats "well" twice before the answer comes to his mind. This repetition does not pose any semantic problems. The ensuing pause is filled with delaying or floor-holding devices. It indicates that the speaker has more to say and does not yet want to yield the floor. The speakers' endeavor is reflected by his employment of "well", which functions as a delay marker. In most of the cases, "well" as a delay marker occurs at the beginning or in the middle of a sentence, and is often accompanied with a short period of silence.

As a Repair Marker
Sometimes, a speaker may make mistakes, and so he/she needs to repair, correct or edit his/her own utterance in the process of his/her current speaking. However, he/she cannot do that abruptly. He/she usually gives the addressee(s) some hints that he/she is going to repair his/her own words. These hints reflect the speakers' endeavor to make the seemingly semantic separated units of talk into coherent ones. "Well" is one of these hints. "Well" of this kind functions as a repair marker, as in (2): (2) A: How fast did you get there? B: I drove ninety miles an hour, well, eight-five all the way to Santa Fe.
In this utterance, Speaker B first states that he/she drove 90 miles an hour, but immediately he/she realized that in the context being concerned a more accurate report of the speech is preferred. Therefore, he/she immediately changes "ninety" to "eighty-five" just after the trouble source. This correction is preceded by the use of the discourse marker "well" to indicate correction or repair. Hence, we label the function being performed by the marker "well" in similar contexts as the repair marker function.

As a Frame Marker
In some other cases the speaker may change or shift his/her present topic to another one for various reasons. He/she may use some signals to indicate that he/she is to change the current topic, and such signals can bring the two separated topics into a more coherent one to avoid an abrupt topic change. "Well" can be such a signal, and function as a frame marker, as in (3) below: (3) A: That will be weird when you're fifty. I wonder what you're like.

B: Nobody's fifty! Don't worry about it.
A: Oh, okay. Oh, well, ten o'clock. Look like it's bedtime for some folks here.
Speaker A and B were talking about getting old in the first two utterances. Then, in the third turn, Speaker A shifted from the previous topic to going bed. This topic shift is accompanied by the use of the discourse marker "well". In this context, the marker "well" is used as a frame marker to separate two topics and indicates a topic change in conversation.

As an Initiation Marker
In addition, "well" can also be used to signal the speaker's acceptance or reception of the addressee's utterance and to claim his/her role to speak as in (4).
(4) Teacher: (when hearing the bell ring) Well, let's begin our class.
In the above situation, "well" serves as a response utterance and initiates a new utterance, thus is labeled as an www.ccsenet.org/ijel International Journal of English Linguistics Vol. 2, No. 5;2012 initiation marker in this paper. Generally speaking, "well" of this use occurs at the very beginning of an utterance, as in (4).
The initiation marker usage and delay marker usage of "well" are, to a large extent, hard to differentiate. In the present study, when the discourse marker "well" is preceded or followed by a silence of more than 2 seconds and is placed in the middle of an utterance, it is categorized as a delay marker; when it is put at the very beginning of the conversation, it is labeled as an initiation marker.

As a Mitigation Marker
"Well" can be used to indicate denials, refusals, and objections as in (5) to signal the speaker's efforts to save others' face.
(5) A: We'll all miss Bill and Marry, won't we?
B: Well, we'll miss Bill.
In (5), "well" is used as a linguistic indicator of politeness, and thus can be regarded as a mitigation marker, which generally occurs at the beginning of a sentence.
There are surely some other pragmatic functions of "well" as a discourse marker, like being an insufficient marker (Ran, 2003), but these functions are quite rare to find in daily conversations. This being so, we omitted these trivial usages and designed the research mainly to investigate how Chinese learners of English and native speakers of English employ those five pragmatic functions of "well" in conversation.

Research Questions
The paper aims to investigate Chinese EFL learners' acquisition of the discourse marker "well" with reference to native English speakers' usages. Specifically, the study attempts to address the following two questions: a) How do Chinese learners of English differ from native speakers of English in the overall occurrence of "well" as a discourse marker? b) How do Chinese learners of English differ from native speakers of English in the use of the pragmatic functions of the discourse marker "well"?

Materials
Two spoken corpora were employed in the study to reveal how native English speakers and Chinese EFL learners use the discourse marker "well" in conversation. The test contains three parts: retelling a story for three minutes; talk based on a given topic for another three minutes plus four minutes' preparation; and a role play lasting for four minutes with a three-minute preparation. The test-takers' performances were recorded in language labs, and then the recorded tapes were mailed to Nanjing University in China for ranking and transcribing into digital forms. Before the transcriptions were put into the corpus, they have been carefully corrected and examined for at least three times to enhance the validity of the transcripts. This study extracted 240 files from the third part (role play for conversation) of the corpus. The sample corpus comprises a total of more than 157 thousand words. The comparison of the two corpora is illustrated in Table1 below:

Data Collection and Analysis
The two corpora were tagged according to the pragmatic functions of "well" illustrated in Section 2 of this paper. Then, the whole size of the each corpus was calculated through the WordList Tool in WordSmith Tools, without counting the tags added by the researcher. Next, the Concord Tool of WordSmith was implemented to find out the general occurrence of "well" as a discourse marker and the respective frequencies of "well" as a delay marker, an initiation marker, a repair marker, a frame marker and a mitigation marker. Thus far, we have got the raw frequencies of the various usages of "well" in the two conversation corpora. As Biber et al. (2000) indicate, raw frequencies should be converted into standardized frequencies in order to make the statistics drawn from the corpora comparable when the sizes of the corpora are different. Thus, standardized frequencies were calculated on the basis of per 10,000 words. When comparing the differences between Chinese learners' and English native speakers' use of the discourse marker "well" and its pragmatic functions, the chi-square analysis was implemented in SPSS 15.0.

Difference on the Overall Frequency
Since the two spoken corpora are different in size, the data presented in this part are all standardized frequencies (per 10,000 words) in order to make a comparable statistical comparison. Note: (1) RF: raw frequency; SF: standardized frequency (per 10,000 words); (2) ***P < .001 It can be seen from Table 2 that Chinese learners of English used the discourse marker "well" 7 times per 10,000 words; while native English speakers' use is far more frequent, with 24 times per 10,000 words. Statistical analysis shows that Chinese speakers significantly underused "well" as a discourse markers in conversation (7 < 24, X 2 = 175.4836, P = .000 < .001).
There is a large amount of evidence that L1 learners begin to have a good mastery of discourse markers in conversation before the age of five or six, and they are likely to be sensitive to sociolinguistic variations of these markers (Sprott, 1992;Kyratzis & Ervin-Tripp, 1993;Anderson et al., 1999). In contrast, the present study, together with He and Xu (2002), reveals that the use of discourse markers like "well" pose great difficulties for both L2 learners. Previous studies have revealed that L2 learners of English are prone to simply repeat the previous word(s), use open silences, or just give up the floor when they are not yet ready for the upcoming turn or when they find great difficulty in expressing themselves (He & Xu, 2003).
Corpus observation shows that Chinese learners of English tend to either use open silences instead of filled ones or use simple interjections like "en", "em", "oh", etc. when the next utterance is still being composed in his/her mind, while the native English speakers are more likely to fill these silences with such discourse markers as "well", "you know", "like", etc.

Difference on the Use of the Pragmatic Functions
From Table 3, we can clearly see that native speakers of English mainly employ the discourse marker "well" as a delay marker (9 times per 10,000 words) and an initiation marker (8 times per 10,000 words). However, Chinese learners of English only frequently employ its initiation marker function in conversation (4 times per 10,000 words), with other functions dramatically lower. Within group analysis shows that both Chinese learners of English and native English speakers prefer to use the discourse marker "well" as a delay marker and an initiation marker. These two functions take at least 65 percent of all the usages for both L2 learners and L1 adult users.  .000*** Note: (1) *** P < .001; (2) For the reason that one marker might have two salient pragmatic functions in the same context, the total frequencies in this table is slightly larger than those in Table 2.
Comparatively speaking, Chinese learners of English significantly underused all the five pragmatic functions of the discourse marker "well", with the sharpest contrast in the delay marker usage (X 2 = 489.6401) and the lowest difference in the repair marker usage (X 2 = 21.663).
Corpus observation shows that when a delay marker is more likely to occur, Chinese learners in most of the cases tend to choose an open silence to compose the forthcoming words; when an initiation marker is necessary, Chinese learners are more likely to start a conversation abruptly, without any indicator of the opening of a new conversation; in context where a repair marker is appropriate, Chinese learners either repeat the trouble source word(s), use non-lexical words like "em", "en", "er", etc, or directly utter the correct form(s); Chinese learners sometimes shift the topic quite incoherently without any indication of the topic shift.
Though Chinese learners are said to be more sensitive to the addressee's face wants (Li & Chen, 2007), the mitigation marker function, which aims to downgrading the impositions of the utterance imposed on the addressee(s), is still among lowest frequency in the conversation corpus. This tells us that we cannot apply a general role to all language learning situations. L2 learners might have quite different strategy preferences in implementing their own actions.

Conclusions
Discourse markers are essential for language production and comprehension (Blakemore, 2002). They are among the most frequently used lexis in conversation (He & Xu, 2003). Among these markers, "well" is, if not ranking first in frequency, among the top ones in L1 conversations. However, few research touch upon the issue of how L2 learners use this marker in conversation. The present paper is designed to fill in this research gap by exploring how Chinese learners use the discourse marker "well" and its pragmatic functions in conversation. The results show that Chinese EFL learners, compared with native English speakers, significantly underused "well" as a discourse marker in general frequency. In addition, they significantly underused the pragmatic functions as well.

Pedagogical Implications
Practically speaking, the study can be applied for pedagogical purposes. As we have detected, discourse markers like "well" are overwhelming in native English speakers' conversations; however, our Chinese learners of English rarely employ such linguistic signals in speaking. Chinese learners' underuse of such discourse markers calls for the following endeavors of textbook compilers, language teachers and language learners: (1) Deeply influenced by the structuralism and TG grammar, our text-book compilers tend to present English learners in China with ideal speakers' utterances which rarely occur in daily communication. Since text-books are the main sources of input from which Chinese learners learn to communicate, as a result, they hardly encounter natural English conversations in their study. In order to tackle this problem, as Leech (2001) proposes, text-book compilers must present with language learners with natural conversations.
(2) Language teachers, in the process of teaching, must raise students' awareness on such small words as discourse markers due to their high occurrence frequency in native speaking. Schmidt (1990) has emphasized the importance of apperception for language acquisition in his Noticing Hypothesis. He proposes that "the subjective experience of noticing is the necessary and sufficient condition for the conversion of input to intake in SLA" (cited in Ellis, 2002: 173). At the same time, language teachers had better explain these markers explicitly since www.ccsenet.org/ijel International Journal of English Linguistics Vol. 2, No. 5;2012 "language acquisition can be speeded by explicit instruction" (Ellis, 2002: 145) and "the explicit types of instruction are more effective than implicit types" (ibid).
(3) Language learners, with exposure to vast English materials nowadays, should pay more attention to how native speakers' utterances are organized and structured. This will facilitate their efforts in later days when they encounter the same communicative needs.

Limitations
A major weakness of the study lies in the classification of the pragmatic functions of the discourse marker "well". The pragmatic functions of the marker "well" are closely related with contexts. The illocutionary force of "well" varies from context to context. This elusiveness adds difficulty to summarize its functions. Therefore, the pragmatic functions of "well" proposed in this paper are surely not exhaustive enough. More efforts are still needed to explain the various functions of it.
Another limitation comes from the materials. Although two spoken corpora, the Santa Barbara Corpus of Spoken American English (SBCSAE) and the Spoken English Corpus of Chinese Learners (SECCCL), were analyzed and compared, we have to admit that the two corpora were built under different conditions. The SBCSAE records daily conversations, whereas the data from SECCL were collected under formal testing environments. Chinese learners' performances in daily conversation might be a little different from those in testing environments.