Effect of L2 exposure: From a perspective of discourse markers

3.1 Participants

This study is a part of a larger project which aimed to examine the influence of four factors on the use of English DMs by Chinese speakers of English (Liu 2009). Ten graduate students from mainland China at an American university participated in this study. Of these, six students (three male and three female) were non-new arrivals who had been studying in the U.S. or Canada for over ten months (1Dong, 2Feng, 3Bing, 4Lian, 5Xia, 6Qiu); the remaining four students (two male and two female) were newcomers who had been in the U.S. for less than a month when the data were recorded (7Peng, 8 Jun, 9Fang, 10Juan). All of them had received formal English education in China for more than eight 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.

Five American L1 English speakers participated in the study (11John, 12Davy, 13Sue, 14Ann, 15Mary). They were graduate students at the same university. They studied in programs such as linguistics, biomedicine, and English.

3.2 Data collection

3.3 Measure of natural exposure of L2

As the research objective of this study is to examine the influence of the Chinese students’ natural exposure to English on their DM use, how to measure their individual amount of exposure is a crucial problem. Milroy’s social network theory is a guide for the current study.

According to Milroy (2002), a network approach has three advantages. First, it provides a method for studying small groups where speakers cannot be grouped into social classes; second, it is a participant-based rather than relying on an analyst’s concept; and, third, it provides a method for dealing with variation amongst individual speakers. In Milroy and Milroy’s (1978) Belfast study, a network strength scale which consisted of a six-point scale going from zero to five was developed which assessed individual speakers’ network characteristics with reference to various relationships within the neighborhood of work and friendship.

The Chinese speakers of English in the current study were assessed according to their self-reported total hours of natural/social English exposure since beginning to study English. The categories that calculated their hours of natural/social English exposure were contact with L1 English-speaking teachers in China, contact with L1 English-speaking friends/ conversation partners/roommates/lab partners/officemates, hours spent watching TV shows (only including natural conversation, e. g., excluding news reports) or movies, class discussion, contact with American advisors, English corner/club, and other situations (depending on individual speakers). The answers were based on their written replies in the questionnaires. The results can be seen in Table 1. I first ordered the ten speakers according to the total hours of exposure to natural/social English that they reported from highest to lowest: speakers 4Xia, 2Feng, 1Dong, 3Lian, 5Qiu, 6Bing, 7Peng, 10Juan, 8Jun, and 9Fang. Thus, I grouped them into two groups: a higher exposure group, in which each speaker had more than 1,000 hours (including speakers 1Dong, 2Feng, 3Lian, 4Xia, and 5Qiu), and a lower exposure group, in which the speakers had less than 1,000 hours, i. e., from 140 to 510 hours of exposure to natural English (including speakers 6Bing, ^[1] 7Peng, 8Jun, 9Fang, and 10Juan).

4.1 English DM use by the L1 English speakers

In the current study, eighteen lexical units were identified as English DMs. Table 2 shows the frequency of DMs used by the five L1 English speakers, in decreasing order of occurrence: and, like, just, y’know, sort of/kind of, I mean, referent-final tags (e. g., and stuff like that, or something, and things like that, etc.), so, I think, well, but, yeah, then, actually, oh, anyway, now, and ah. As can be seen from the table, an average of 71.5 DMs per 1,000 words was used by the L1 speakers. And, like, just, y’know, sort of/kind of, and I mean have the highest rates of occurrence. Individual differences among the five speakers can also be seen in the table.

4.2 English DM use by the Chinese speakers of English

Table 3 shows the use of English DMs by the ten L2 English speakers in decreasing order of frequency: yeah/yes, and, I think, y’know, ah, just, like, oh, so, sort of/kind of, actually, referent-final tags, well, but, now, I mean, then, and anyway. As can be seen, the most frequently used markers by the L2 English speakers are yeah/yes, and, and I think. These markers were used by every speaker. The rest of the markers, which have lower frequencies, were not used by all the speakers, and none of the L2 English speakers used then, or anyway. An average of 66.5 DMs per 1,000 words was used by the L2 English speakers.

4.3 The higher exposure group vs. the lower exposure group

In order to examine the influence of natural/social exposure to English on the English DM use by the Chinese speakers of English, I grouped them into two groups (see discussion in Methodology Section). In order to assess the different use of English DMs, the occurrences per 1,000 words in the data for the higher exposure group and the lower exposure group are presented in Table 4. Results of Independent-Samples T tests can also be found in Table 4. As can be seen, the higher exposure group used 74.0 markers per 1,000 words per person, while the lower exposure group used 60.6 markers per 1,000 words per person. The higher exposure group used yeah/yes, I think, y’know, like, oh, so, actually, sort of/kind of, referent-final tags, well, but, now, and I mean at higher rates than the lower exposure group. The lower exposure group used and, just, and ah at higher rates than the higher exposure group. None of the speakers in the higher exposure group used anyway, or then. None of the speakers in the lower exposure group used like, but, now, I mean, anyway, or then. None of the DMs has been found to be used differently at statistically significant level (p=.05) by the two exposure groups.

To sum up, there was a tendency for the higher exposure group to exceed the lower exposure group in both frequency and variety of DMs.

5.1 And

Among the most prevalent six markers used by the L1 English speakers, only and was used by both the higher exposure group and lower exposure group at similarly high rates as the L1 speakers. As for the other five markers, both exposure groups used them at much lower rates than the L1 speakers. It seems that the DM and was easier for the L2 English speakers to acquire than other markers.

In my data, and serves a textual function—to hold the floor and continue the speaker’s thoughts. The L1 speakers and the speakers of the two exposure groups used and for the same function, shown in the following examples by Mary, Xia, and Juan.

5.2 Like

Like was the second most frequently used DM by the L1 English speakers (8.7 tokens per 1,000 words per person). None of the five speakers in the lower exposure group used the DM like in the data. The higher exposure group used 4.1 tokens per 1,000 words per person; the p value of like in Table 5 also suggests that the difference between the two exposure groups neared significance (p = 0.069).

Schourup (1985) and Andersen (1998) suggest that a primary function of the DM like is to indicate approximation. Underhill (1988) claims that the function of like is as a focus particle, serving to highlight a speaker’s point in an utterance. Müller (2005: 204) summarizes four textual functions of DM use of like: “searching for the appropriate expression”, “marking an approximate number or quantity”, “introducing an example”, “introducing an explanation”, and “marking lexical focus”. My study will follow Müller’s analysis. In (4), the underlined likes are used to gain time to search for the appropriate expression. When used as a delay device, like is often followed by a pause or in collocation with another DM, such as y’know in this example.

In (5), the underlined like marks an approximate number or quantity. Like preceding a numeral expression indicates that the number is not the exact one (Müller 2005).

In (6), the speaker introduces the topics that a class covered. The underlined like is used as an exemplifier. ^[4] Here examples are not preceded by the explicit mention of a general concept – topics covered in that class. The speaker has such a concept in mind, but only expresses the examples (Müller 2005). In this situation, it is syntactically optional.

In (7), the underlined like introduces an explanation. Sue was trying to explain “happy hour” to the interviewer. The speaker extended the information after like in order to make it more understandable for her partner.

The last function of like in the data is a marker of new information and focus. Focus is “the most significant new information in a sentence – often, the point of the sentence” (Underhill 1988: 238). In (8), the underlined likes were used to mark new and important information by the speaker.

In order to compare the use of sub-functions of like between L1 and L2 speakers (including two exposure groups), I calculated the tokens and percentages of sub-functions of like respectively and the results are presented in Table 5. We can see that the L1 English speakers used ‘searching for the appropriate expression’ function and ‘marking lexical focus’ function at the highest rates. The higher exposure group also used these two functions at the highest rates. However, they were not found to use ‘marking an approximate number or quantity’ function of like in the data. The reason could be that even though the higher exposure group acquired the use of like when they communicated with L1 speakers of English, they had not yet acquired the ‘marking an approximate number or quantity’ function.

Four out of the five L1 speakers and four out of the five L2 higher exposure group speakers used like in the data. The exceptions were an L1 participant Ann and an L2 participant Xia. Ann ^[5] was an English major graduate student; Xia was an English major in China and a linguistics graduate student in the U.S. Both of them were found to use the total number of DMs at the lowest rates among their groups respectively (see Tables 2 and 3). After the data analysis, I interviewed Xia about the use of the DM like. Xia said that she avoided using like on purpose whenever she spoke English, because she thought that the marker was stigmatized and she preferred a more formal English style. The reason why Xia was different from the other speakers in the higher exposure group may be that her training in the English major made her consciously choose a prescriptive speech style, while the other speakers either did not care about it very much or their English proficiency was not high enough for them to notice the degree of formality of English, and they subconsciously imitated the oral speech of the L1 English speakers around them.

5.3 Just

Just was the third most frequently used DM by the L1 English speakers (8.5 tokens per 1,000 words per person), at a much higher rate than the L2 speakers (2.2 tokens per 1,000 words per person). The higher exposure group used just at a similar rate (2.1 tokens per 1,000 words per person) as the lower exposure group (2.3 tokens per 1,000 words per person). Every speaker in the higher exposure group was found to use this marker, however only two out of the five speakers in the lower exposure used it.

Just as a DM has two main interpersonal functions. When it is an emphatic/ strengthening/intensifying marker, just is a strategy of positive politeness; when it is used as a weakening/downtoning hedge, it signals negative politeness (Aijmer 2002). The two functions are shown in examples (9) and (10) respectively.

Table 6 shows the tokens and percentages of sub-functions of just by the L1 speakers and two exposure groups. The L1 speakers used intensifying function of just at a similar rate to downtoning function. The higher exposure group used just mainly for downtoning hedge. The lower exposure group used the intensifying function at a higher rate than the higher exposure group.

5.4 Y’know

Y’know was the fourth most frequently used DM by the L1 English speakers (6.8 tokens per 1,000 words per person). Four out of the five speakers in the higher exposure group used y’know in the data (4.5 tokens per 1,000 words per person); only one out of the five speakers (Peng) in the lower exposure group used it (3.5 tokens per 1,000 words per person).

Müller (2005: 157) summarizes previous works and lists five textual functions: “marking lexical or content search”, “marking false start and repair”, “marking approximation”, “introducing an explanation”, and “introducing a quotative y’know”.

Y’know is found to have four textual functions in my data. The first textual function marks a lexical or content search. Usually it is followed or preceded by a pause. For example, in (11), the underlined y’know is preceded by one “uh” and two “that”s and followed by one “that” which all indicated that the speaker was searching for the appropriate expression here.

The second textual function marks false start and repair as in (12). John was answering the interviewer’s question about why he didn’t watch TV. He made a false start at the beginning of his turn. After “y’know,” he corrected his false start and said that he had flipped some channels at first but turned it off later.

The third textual function introduces an explanation as in (13). Davy was talking about his travels in Canada. Before “y’know,” he said that traveling was pretty much the same in Montreal as in other places. After “y’know,” he explained in more details what he did there.

The fourth textual function involves quotative y’know. In (14), the speaker was telling the interviewer why he chose UF as his graduate school. The underlined y’know is followed by a quotation in this case. ^[6]

Besides the textual functions, y’know is found to have an interpersonal function, as in example (15). It is used to presume the common ground between speaker and hearer (Schourup, 1985).

Table 7 presents the tokens and percentages of sub-functions of y’know by the L1 speakers and the two exposure groups. The L1 speakers used all five functions of y’know. The higher exposure group was found to use four of the five functions but not ‘introducing quotative y’know’ function. The lower exposure group used three of the five functions but not ‘introducing quotative y’know’ and ‘marking false start and repair’ functions. Both the L1 and two exposure groups used ‘marking lexical or content search’ function at the highest rate among the five functions.

Hays (1992) mentions that y’know would not be expected to be used frequently by L2 English speakers unless learners are exposed to the target language discourse community. In my study, however, Peng in the lower exposure group used y’know at a high rate in the data, while the other four speakers in the lower exposure group did not use it at all. Peng used y’know at an even higher rate than the other speakers in the higher exposure group (see Table 3). He had been in the U.S. only two weeks when the data were collected. After the data analysis was finished, I interviewed him about his use of the marker y’know. He said he was a highly motivated English learner when he was in China. He even spoke English with his father at home and acquired this marker from his father. He said his father’s English was good because his father had been working in Germany for several years. Thus although he was in the lower exposure group due to his short length of residence in the U.S., his high motivation to learn English led to his grasping the learning opportunities in his environment efficiently.

5.5 Sort of/kind of

The two markers are interchangeable in this study. Aijmer (2002: 209) describes them as ‘hedging particles’ because they are used when speakers avoid being precise. This interpersonal function of sort of/kind of is “closely associated with claiming common ground and avoiding disagreement” (Aijmer 2002: 191). In (15), Sherry asks Davy what his favorite movie is. Davy thinks it is a hard question to answer. But by using kind of, he avoids being precise, thus distancing himself from the responsibility for describing the interviewer’s question as hard.

Sort of/kind of was the fifth most frequently used DM by the L1 English speakers (6.7 tokens per 1,000 words per person). All the L1 speakers used this marker, but only half of the L2 speakers were found to use it in the data. The higher exposure group used it at a slightly higher rate (1.6 tokens per 1,000 words per person) than the lower exposure group (1.1 tokens per 1,000 words per person). Both groups (Feng and Bing in examples 16 and 17) used the marker for the same function as the L1 speakers.

5.6 I mean

Schiffrin (1987: 296) suggests that I mean marks “a speaker’s upcoming modification of the meaning of his/her own prior talk”. Tree and Schrock (2002) find that I mean may have an interpersonal function: I mean may be linked with positive politeness, because using DMs reminds the interlocutors of more casual talk, and thus indicates intimacy of their relationship; and it may also be linked with negative politeness, because I mean reduces the degree of the speaker’s commitment to what s/he just said and s/he adjusts the utterance.

In my data, I mean occurs utterance-initially and medially but not finally. It is found to have a textual function – introducing an explanation or justification. In (18), the speaker talks about the TV show he likes. He uses I mean to start elaborating on reasons for his choice. I mean has an interpersonal function as well. For instance, in (18), on the one hand, I mean reduces the formality of the talk between interlocutors; on the other hand, using I mean decreases the degree of the speaker’s commitment to what he just said in the previous part of the utterance. This interpersonal function does not contradict the textual function.

I mean was the sixth most frequently used DM by the L1 English speakers (5.6 tokens per 1,000 words per person). Three out of the five L1 speakers used this marker in the data. The L2 speakers used I mean at a much lower rate of 0.1 tokens per 1,000 words per person. Out of the ten L2 speakers, only one speaker (Dong) from the higher exposure group was found to use this marker once in the data. He used this marker for the same function as the L1 speakers, as shown in (19).