The use of refusal strategies in interlanguage speech act performance of Korean and Norwegian users of English

The present paper describes a contrastive study of interlanguage refusal strategies employed by Korean and Norwegian learners of English as an additional language. The data were collected from multilingual first-year students at an American university in South Korea and in an English-medium program at a Norwegian university by means of an online open discourse completion task and analyzed using the coding categories based on Beebe, Takahashi, and Uliss-Welts (1990), and Salazar Campillo, Safont-Jordà, and Codina Espurz (2009). The data were analyzed to compare the average frequencies of refusal strategies used by the two groups, and the types of direct, indirect, and adjunct strategies that they employed. Independent samples t-tests revealed significant differences in the use of direct and indirect strategies with small effect sizes. The differences in the use of adjunct strategies were not statistically significant, and the effect sizes were negligible. Descriptive statistics of the differences in the types of direct, indirect, and adjunct strategies also revealed interesting patterns. The findings suggest that multilinguals’ pragmatic Anna Krulatz, Tülay Dixon 752 performance is a complex phenomenon that cannot be explained by the differences in cultural and pragmatic norms of their first language alone.


Introduction
The ability to produce speech acts is a subcomponent of sociolinguistic competence, which accounts for the knowledge of the sociocultural rules that govern language use (Canale & Swain, 1980). Previous research suggests that due to varying social norms among speech communities, performance of speech acts is not uniform across different cultures (e.g., House & Kasper, 1981;Wolfson, 1989) and that increased proficiency in the target language (TL) is not a reliable predictor of native-like performance (Xiao, 2015).
Refusals are typically formulated in response to a range of other speech acts, including invitations, requests, and suggestions, and entail a possibility to offend or upset the interlocutor. In fact, it has been suggested that refusals pose a face-threat to both parties involved in the interactional exchange (cf. Johnson, Roloff, & Riffee, 2009). The person making a refusal violates relational expectations of the requester, especially if this person has a high ability to comply, while the person receiving a refusal risks coming across as incompetent to select a person with a high ability to comply. To mitigate the offence -and to save face -refusals often consist of a long and complex sequence of semantic formulas. The choice of strategies used to perform a refusal is dictated by the contextual variables, such as the status of the interlocutor and the level of imposition. Consequently, performance of refusals requires selection of appropriate linguistic means and access to extensive cultural knowledge and can thus pose a challenge to adult language learners (Nelson, Carson, Al Batal, & El Bakary, 2002;Salazar Campillo, Safont-Jordà, & Codina Espurz, 2009).
To date, research on interlanguage refusals has examined performance by English learners of various first language backgrounds. However, to our knowledge, while some studies have compared English refusals produced by Korean learners of English with those produced by native speakers (e.g., Jung & Kim, 2008;Kwon, 2004;Lee, 2013;Lyuh, 1994), no previous research has investigated refusals by Norwegian learners of English, and only one previous study has presented a limited qualitative analysis of interlanguage refusals by Korean and Norwegian speakers of English (Krulatz & Dixon, 2016). The reason these two groups are of particular interest is because the Korean and Norwegian cultures differ substantially on cultural dimensions of power distance and individualism (Hofstede, 2001), which may have an impact on how their members produce face-threatening acts such as refusals. The Korean culture is believed to be somewhat hierarchical and highly collectivist, while the Norwegian culture is highly egalitarian and individualist, implying the possibility of refusal strategy difference related to directness and status of the interlocutor. Although Hofstede's framework has been criticized on several grounds (Jones, 2007;McSweeney, 2002;Signorini, Wiesemes, & Murphy, 2009), his approach remains to be commonly employed in the research on cultural differences (Gong, Lee, & Stump, 2007;Holden, 2004). We want to remain cautious, however, about the possibility of making inaccurate cultural generalizations and acknowledge that it is important to recognize intra-language variation present in any cultural group.
Research suggests that while pragmatic norms vary across languages and cultures, multilinguals have enhanced pragmatics awareness and increased flexibility in the selection of appropriate linguistic choices in comparison with monolinguals (Cenoz & Hoffman, 2003;Kecskés, 2015;Safont-Jordà, 2005). However, most research on interlanguage to date has adopted the construct of a native speaker as a point of departure and compared non-native performance to an estimated monolingual native-speaker norm (Cenoz & Gorter, 2014). This is problematic not only because pragmatic norms are arguably impossible to establish but also because the monolingual native speakers of English are a minority group among the users of English worldwide, with multilingual speakers being the norm rather than exception (Cenoz & Hoffman, 2003;Cook, 2009;Grosjean, 2010). For these reasons, the present study focuses on multilingual speakers of English. We take the position that non-native use of English by multilinguals of different linguistic backgrounds is an important area for pragmatics research. Following current dominant perspectives on bilingualism and multilingualism, we adopt a broad definition of multilingualism that does not require balanced proficiency in all languages (Aronin & Singleton, 2012). Thus, for the purpose of this article, multilingualism is defined as "the ability to use languages for the purposes of communication and to take part in intercultural interaction, where a person . . . has proficiency of varying degrees, in several languages, and experience of several cultures" (Council of Europe, 2001, p. 168 Hofstede's (2001) dimensions that label national cultural differences (i.e., power distance, individualism, masculinity, uncertainty avoidance, and long-term vs. short-term orientation) can be helpful in explaining the reasons why different cultures realize certain speech acts differently. For refusals, it is important to consider the power distance dimension, which refers to the inequalities among members of a society (Hofstede, 2001), as the inequality -or equality -among its members can impact to what extent and using what means a refusal is mitigated. According to Hofstede (2001), South Korea and Norway are substantially different in terms of power distance, with South Korea scoring 60 and Norway 31 out of 100. This means that South Korean society is somewhat hierarchical while Norwegian society is egalitarian. Consequently, unlike in Korea, markers of social distance in Norway tend to be eliminated from discourse to underscore equality between interlocutors and to minimize the level of imposition (Awedyk, 2003;Dittrich, Johansen, & Kulinskaya, 2011;Fairclough, 1992;Rygg, 2012). Given that Koreans are more status-conscious than Norwegians, and that the differences in social status influence the selection of linguistic means in performance of refusals, it can be anticipated that Koreans are likely to use more extensive mitigation strategies to minimize threats to the interlocutor's face, in particular when communicating with an interlocutor of higher social status, both in their first language and additional languages (e.g., English). On the other hand, Norwegians are expected to employ fewer strategies to soften their refusals.
Another cultural dimension relevant for the study of speech acts is individualism, which explains the degree of interdependence among the members of a society (Hofstede, 2001). South Korea scores fairly low on this dimension (18), with Norway's score being relatively high (69). Thus, South Korea is considered a highly collectivistic society in which members have a strong group commitment and assume responsibility for fellow group members. Loyalty is an important value, and an offence leads to shame and loss of face. Norway, on the contrary, is a highly individualist society, in which personal opinions are respected and freely expressed, and the right to privacy is highly valued. It can therefore be expected that there are differences in the length and the content of refusal strategies employed by Korean and Norwegian speakers of English, namely, that the Korean participants employ longer strings of strategies and more indirect strategies than the Norwegian participants to minimize the imposition on the interlocutor.
We fully acknowledge that Hofstede's (2001) framework has been criticized on several grounds, including his methodology (Jones, 2007;McSweeney, 2002), the use of nations as a basis for measuring cultural aspects (McSweeney, 2002), the oversimplification of cultural differences, the assumption that cultural values are stable, and some inconsistencies between the categories in the framework (cf. Signorini et al., 2009). Nevertheless, Hofstede's conclusions have not been successfully discredited, and his approach remains to be the most commonly recognized, robust and, comprehensive framework employed in research on cultural differences (Gong et al., 2007;Holden, 2004).

Refusals in Korean and Norwegian
Refusals in Korean have been studied by Lyuh (1994) and Kwon (2004), both of whom compared the refusal strategies employed by Korean speakers and American English speakers. The results of these studies indicate that Koreans employ more complex semantic formulas and less direct strategies in their first language (L1) than Americans. When refusing someone of a higher status, Korean speakers, but not English speakers, use extensive mitigation. Direct refusals are used less frequently by Korean speakers in comparison with English speakers, and reasons (an indirect strategy) are often employed by the Koreans as excuses for refusals, which could be attributed to the fact that direct disagreement is often considered hostile in Korea's collectivist society (Jung & Kim, 2008). In addition, Korean speakers typically apologize before refusing, while English speakers show gratitude or express positive opinions. In all, Korean refusals tend to come across as less direct and transparent and more tentative than English refusals.
No study to date has investigated the speech act of refusal in Norwegian. A general, baseline study of stereotyped politeness phrases in Norwegian was undertaken by White (1979), who described basic greetings, responses to greetings, farewells, requests, apologies, expressions of gratitude, and responses to expressions of gratitude. White's (1979) paper, however, contains a mere list of politeness expressions collected via "eavesdropping" (p. 3), accompanied by explanations of when these expressions should be used. Regarding apologies, which are similar to refusals as they entail complex semantic formulas and can be used as adjuncts to refusals, White (1979) lists expressions such as unnskyld 'sorry,' unnskyld meg 'excuse me,' beklager 'sorry,' desverre 'unfortunately,' and jeg er lei meg 'I am sorry,' and provides examples of contexts when these could be used. She also elaborates that in situations when Norwegians want to offer a sincere apology, they "often find some more original way of expressing their sympathy or pleasure . . . in order to indicate real sincerity rather than merely dutiful politeness" (White, 1979, p. 40). These claims, however, are not based on any empirical evidence and can therefore only be taken as hearsay.
Other relevant research on politeness and other speech acts such as requests suggests that Norwegians use conversational strategies that stress equality between interlocutors. Awedyk (2003) found a large degree of similarity between the requestive strategies used by his English-speaking and Norwegian-speaking subjects. A study by Dittrich et al. (2011) compared how politeness is performed in Norway and Great Britain by examining the forms of address employed in equal and unequal relations with others, and found a higher prevalence of title use among the British subjects, while the Norwegian participants opted for more personal and informal usage in all situations. We now move on to what previous research has found regarding refusals by Korean and Norwegian speakers of English.

Refusals by Korean and Norwegian learners of English
The existing research on refusals of Korean learners of English suggests that although Korean English speakers and native English speakers employ some of the same refusal strategies, the frequency and content of these strategies are different. In comparison with native speakers, Korean speakers of English tend to use fewer direct refusals (Jung & Kim 2008;Lyuh, 1994), which could be attributed to transfer from Korean, as the same pattern observed in native Korean performance (Kwon, 2004). In terms of the content of the reasons that are given as part of the refusal, Lyuh (1994) noted that Korean speakers in his study referred to existential circumstances that prevented them from accepting a request, whereas English speakers' reasons were more personal. Jung and Kim (2008) and Krulatz and Dixon (2016) also noted that Korean speakers' reasons were more elaborate than reasons provided by native English speakers. Another difference between the two groups relates to the strategies used before the head act of refusal itself. Korean English speakers often apologize or show regret before refusing (Jung & Kim, 2008), which has been observed in the refusals of native speakers of Korean as well (Kwon, 2004).
In Lee's (2013) study, the status difference was also observed found to affect the fluency level of Korean English speakers. The participants took more production time when refusing someone of a lower status. The results are unanticipated as it is usually expected that refusing someone of a higher status would be more difficult and, thus, take more production time. According to Lee (2013), Korea is going through cultural changes which involve moving from a collectivist society to an individualistic one. Thus, maintaining self-image is becoming more important than saving other people's face. Lee (2013) suggests that as refusing someone of a lower status might reflect negatively on a person's public image, Korean English speakers take more time to do so.
Familiarity with interlocutors is yet another variable that affects how the speech act of refusal is realized. Jung and Kim (2008) accounted for this variable when they compared the refusal strategies employed by Korean and native speakers of English. Their results showed that the use of direct strategies increased for both groups when they were refusing familiar people in comparison to refusing unfamiliar people. As for indirect strategies, Korean English speakers used more indirect strategies with unfamiliar interlocutors while native speakers did so with familiar interlocutors. With familiar interlocutors, Korean speakers also resorted to more adjunct strategies (i.e., positive feeling, empathy, pause fillers, and gratitude) compared to native English speakers.
Besides the baseline report by Krulatz and Dixon (2016), to our best knowledge, no other study has examined refusals by Norwegian users of English. Krulatz and Dixon (2016), who used some of the same data analyzed in the current study, examined refusals by Korean and Norwegian speakers of English and found that both groups have difficulties pertaining to the use of main refusal strategies (i.e., direct vs. indirect), and overuse, underuse, or misuse of indirect and adjunct strategies (e.g., advice, criticism, gratitude, and reason). However, this study only used a limited data sample and did not include an in-depth analysis of refusal strategies.
A few other studies have explored performance of other speech acts by Norwegian speakers of English, including requests (Brubaek, 2012;Krulatz, 2016;Savić, 2015) and expressions of gratitude (Johansen, 2008). Brubaek (2012) investigated English requests by Norwegian high school students focusing on their ability to adapt choice of strategy and the level of formality to contextual demands. She concluded that the participants transferred request strategies from their first language and overused familiar and informal expressions. Krulatz (2016) analyzed request strategies employed by Norwegian teachers of English in scenarios with varying social distance. She found that these advanced users of English employed a range of requestive strategies and varied their requestive behavior depending on the social distance between the interlocutors. In her cross-sectional study, Savić (2015) examined the development of requests in young Norwegian learners and reported clear patterns of pragmalinguistic development, in particular the complexity of head acts and the use of alerters, supportive moves, and downgraders. Finally, Johansen (2008), who compared expressions of gratitude by native speakers of English and Norwegian learners of English, reported differences between how these two groups interpreted the social distance and the size of the imposition in their choices of negative and positive politeness strategies.

The multilingual orientation in the study of pragmatics
With the increased globalization and mobility, multilingualism is becoming more ubiquitous, and it has been recognized as the new linguistic dispensation that has replaced the monolingual norm (cf. Aronin & Singleton, 2012;Stavans & Hoffman, 2015). However, with some exceptions (e.g., Safont-Jordà, 2013;Stavans & Webman Shafran, 2018), the majority of studies on pragmatics in additional languages continue to focus on learners' deficiency in pragmatic competence, as measured by the monolingual yardstick, and ignore their multilingual backgrounds. As with other components of communicative competence, negative transfer from the first language has traditionally been identified as a source of pragmatic failure (Thomas, 1983), and the challenge of attaining native-like pragmatic competence has been attributed to cross-cultural and cross-linguistic variation in pragmatic norms (Rose & Kasper, 2001). By contrast, research on pragmatic performance through the multilingual lens suggests that, compared with monolinguals, multilinguals display enhanced pragmalinguistic awareness and increased flexibility in the selection of appropriate linguistic choices (Cenoz & Hoffman, 2003;Kecskés, 2015;Safont-Jordà, 2005). In the present study, we move away from the deficiency view of pragmatic competence in learners of English as an additional language and, rather than comparing native and nonnative performance, we examine two groups of multilingual learners of English.
Recognizing that balanced pragmatic competence in several languages is an unrealistic goal (Cenoz, 2007), we consider these language users in their own right.

Research questions
To date, research on interlanguage refusals by multilingual learners of English has been limited. At the same time, Korean and Norwegian learners of English constitute interesting comparison groups considering the large discrepancies in the scores on Hofstede's (2001) dimensions of individualism and power distance. The present study aims to contribute to this relatively neglected area of interlanguage pragmatics research by exploring the following research questions:

Participants
Participants in this study consisted of two groups: 81 Korean college students, native speakers of Korean (age: 18-35, males: 31, females: 50), and 62 Norwegian college students, native speakers of Norwegian (age: 18-35, males: 13, females: 49). All participants were enrolled in English-medium courses at an institution of higher education in their country of origin. The majority of the participants (91% of the Korean informants, and 76% of the Norwegian informants) reported that they had resided in an English-speaking country prior to participation in the study. The length of residence is summarized in Table 1. The participants were asked to report their proficiency (basic, intermediate, or advanced) in other languages besides their first language and English. All of them admitted some level of proficiency in at least one additional language: 54% in one additional language, 33% in two additional languages, and 13% in three or more additional languages. Thus, as per the definition of multilingualism adopted in this study, all participants can be considered multilinguals. The participants were also asked to self-report how comfortable they felt regarding their ability to communicate with native speakers of English. Most Korean respondents (85%) and Norwegian respondents (92%) answered that they felt comfortable in all or most situations or environments. These data are summarized in Table 2.

Data collection procedures
Data were collected through an online discourse completion task (DCT) consisting of two refusal scenarios, which required participants to write email refusals. A link to an online survey containing a background questionnaire and a discourse completion test consisting of two refusal scenarios was sent to the participants via email (see Appendix). The survey was administered in English, and the participants had unlimited time to respond. The responses were fully anonymous, and no IP addresses were stored.
Only two DCTs were used in the survey to reduce the serial task effect. Although an important limitation of DCTs is that responses may not correspond to how participants would react in natural settings, this elicitation method is commonly used in interlanguage pragmatics research as it allows collection of large samples of learner data containing the speech act in focus as well as control over different variables such as social distance and power dominance (Chaudron, 2003;Jucker, 2009). Additionally, requiring participants to write emails enhanced the authenticity of the task.
The first scenario elicited a refusal in response to a peer's invitation to his birthday party (an equal status refusal), and the second one targeted a student's refusal to a professor's request to give a short talk (a different status refusal). Consequently, the two scenarios differed with regard to the parameters of distance and power (cf. Blum-Kulka, House, & Casper, 1989), as summarized in Table  3. These parameters were varied to ensure elicitation of varied request strategies while at the same time limiting the number of scenarios to two helped avoid participant fatigue.

Data analysis
Salazar Campillo et al. (2009) propose a taxonomy for refusal analysis that is based on Beebe et al. (1990). Both taxonomies divide refusal strategies into three categories: direct strategies, indirect strategies, and adjuncts to refusals; however, the strategies under these categories differ slightly. In the present study, the taxonomy of refusals is based mainly on Salazar Campillo et al. (2009) but complemented with some of the strategies from Beebe et al. (1990). Table 4   Slight changes were made to the taxonomy by Salazar Campillo et al.
due to the type and nature of the data collected. Wish was found in the data but could not be accounted for using the strategies outlined in Salazar Campillo et al.'s (2009) coding taxonomy, and thus was added from Beebe et al.'s (1990) taxonomy. In addition, as the data in this study were comprised of written e-mails, the non-verbal strategies under the avoidance category were removed.
The distinction between a reason and statement of principle was unclear in sentences such as "I am not a person who likes to stand in front of people," "I don't feel comfortable standing in front of an audience," and "I do not feel confident to teach others." Such sentences were marked as statements of principle because they refer to "reasons" that are permanent, which makes them part of the participants' disposition and, thus, statements of principle. An example of a reason is "I have previous arrangements." Another challenge was coding refusals preceded by unfortunately, sadly, and regrettably, as in "Unfortunately I can't go to your party." In this example, the word unfortunately denotes regret; however, coding the whole statement as regret is problematic, as unfortunately is followed by a negation of proposition. In such cases, unfortunately, as well as other similar words that denote regret (e.g., sadly), was marked as a hedge.
The difference between the positive opinion and gratitude strategies should also be clarified. Gratitude refers to comments of appreciation and thankfulness only for the proposition itself (e.g., Thanks for inviting me to the party), whereas positive opinion refers to comments in which the participants find the proposition to be a good idea (e.g., I'm sure the students would greatly benefit from this, but . . .). However, the territory of the positive opinion category was expanded due to the refusal scenarios in this study. One scenario asked the participants to refuse a friend's birthday party invitation, while the other asked them to decline their professor's request to do an in-class presentation on a paper that the participants wrote in a previous semester. In order to accommodate these scenarios, the positive opinion category included statements in which the participants congratulated friends on their birthday and appreciated the professor's compliment on their paper (e.g., "I hope you have a blast at your party and happy early birthday" or "Thank you for your praise about my paper").
Before coding the data, the researchers went through three cycles of training on sample refusals and jointly discussed any problematic or challenging codes. Next, the data were coded independently by the two researchers using the codes agreed upon (Beebe et al., 1990;Salazar Campillo et al., 2009). Each refusal was divided into its component strategies, and each strategy was matched with a code. For example, the refusal by a Norwegian participant below was divided and coded as follows: ( The simple percentage agreement was used to calculate interrater reliability because the coding scheme was not dichotomous. Interrater reliability was 92%. The researchers then discussed and resolved any discrepancies together with the help of the coding guidelines until they reached a consensus. A total of 286 refusals were produced by the participants in response to the DCTs in this study. However, after the removal of infelicitous responses (i.e., acceptance of invitation or request), 157 refusals by Korean participants and 122 refusals by Norwegian participants (a total of 279) were included in the analysis. The analysis involved calculating: (a) the frequency of strategy use and (b) the types of direct, indirect, and adjunct strategies used by the two groups.
The differences between groups were analyzed using descriptive statistics (means and standard deviations), independent-samples t-tests, and effect sizes. For t-tests, the data were checked for normality using histograms and Q-Q plots and were deemed to be normally distributed. The alpha level was set to .05 for all statistical analyses. The effect sizes were interpreted using the following benchmarks suggested for L2 research: d values around 0.40 as a small effect size, values around 0.70 as a medium effect size, and values around 1.00 as a large effect size (Plonsky & Oswald, 2014).

Frequency of strategies used
To answer research question 1, the means of total strategies per item as well as the means of total direct strategies, indirect strategies, and adjunct strategies for each of the two groups were compared (see Table 5). On average, the Korean participants utilized more refusal strategies (5.17 strategies per item) than the Norwegian participants (4.61 strategies per item). However, this difference was found to be statistically non-significant on the basis of an independent samples t-test for mean differences (t(277) = 2.737, p = .07). In addition, the effect size for this analysis was small, d = 0.33. On average, both the Korean and the Norwegian participants utilized substantially more indirect than direct refusal strategies, as shown in Table 5. The Norwegian participants were observed to use direct strategies more frequently than the Korean participants (0.69 vs. 0.5 direct strategies per item), a difference which was found to be statistically significant with a small effect size (t(277) = -3.084, p = .002, d = 0.37). However, the Korean participants employed, on average, more indirect strategies than the Norwegian participants (2.82 vs. 2.31 strategies per item). This difference was also found to be statistically significant with a small effect size (t(277) = 3.405, p = .001, d = 0.41). However, no significant difference or large effect were found in the average number of adjunct strategies per item (t(277) = 1.607, p > .05, d = 0.19).
A typical example of a Korean refusal consisting of direct, indirect, and adjunct strategies is illustrated in (2)  As can be seen in Table 6, the same patterns in the differences between the two groups were observed in the responses to the two refusal situations: higher status (professor) and equal status (peer; research question 2). The Korean participants used on average more strategies per item both in the higher and equal status scenario. The differences between the groups were statistically significant, but the effect sizes were small (higher status: t(135) = 2.026, p = .045, d = 0.35; equal status: t(140) = 1.975, p = .05, d = 0.32). However, although the Korean participants employed on average more adjunct strategies per item than the Norwegian participants, these differences failed to reach statistical significance, and the effect sizes were negligible (higher status: t(135) = 1.755, p = .082, d = 0.3; equal status: t(140) = 0.900, p = .37, d = 0.16).
Statistically significant differences were observed in the use of direct and indirect strategies in the two refusal situations. In response to the higher status scenario, the Norwegian participants employed on average more direct strategies (0.52 per item) than the Korean participants (0.30 per item), a statistically significant difference with a small effect size (t(135) = -2.607, p = .01, d = 0.46). The Korean participants, on the other hand, employed on average more indirect strategies (3.01 per item) than the Norwegian participants (2.52 per item) (t(135) = 2.251, p = .03, d = 0.38). Likewise, in the equal status scenario, the Norwegian participants showed a slight preference for direct strategies (0.85 per item) as compared with the Korean participants (0.69 per item), t(140) = -2.051, p = .04, d = 0.33. By contrast, the Korean participants used on average more indirect strategies (2.64 per item) than the Norwegian participants (2.11 per item), t(140) = 2.628, p = .01. This difference had a small to medium effect size (d = 0.62).

Types of direct, indirect, and adjunct strategies
The use of direct, indirect and adjunct strategies was examined to answer research question 3 ("Do Korean and Norwegian learners of English use different direct, indirect, and adjunct strategies?"), As shown in Figures 1-3, both groups of participants used primarily one type of direct strategies (negation of proposition), five types of indirect strategies (reason, regret/apology, alternative, hedging, and plain indirect), and three types of adjuncts (positive opinion, gratitude, and willingness). Inferential statistical comparisons were not made due to the large number of variables (nine strategy types, two different scenarios, and two groups). However, descriptive statistics together with qualitative analyses reveal some interesting patterns of strategy use by the two groups, and these are illustrated in the sections that follow with examples from the data.

Direct strategies
Direct strategies accounted for 14.9% of all strategies used by Norwegians and 9.4% of all strategies used by Koreans. Both types of direct strategies from the coding taxonomy, that is, negation of proposition and bluntness, were present in the data. As Figure 1 illustrates, negation of proposition was utilized somewhat more frequently by the Norwegian participants (11.9% of all strategies) than by the Korean participants (9.4%), whereas bluntness was only employed by the Norwegian participants (3.0%).

Figure 1 Frequency of direct refusal strategies
Both groups tended to provide a reason for refusal immediately after negating a proposition, as illustrated in (4) When employed, bluntness also tended to be accompanied by other mitigating strategies such as reasons, apologies, and hedging. Two instances of minimal or no mitigation of bluntness by Norwegian participants stand out, as illustrated in (5)

Indirect strategies
Overall, indirect strategies were the most frequently employed strategies in the data, and these strategies were used in a similar fashion in both equal and unequal social status scenarios. They constituted 49.9% of all strategies used by the Norwegian participants, and 54.8% of all strategies used by the Korean participants. Within these, reason and regret/apology were used the most often, followed by alternative, hedging, and plain indirect (Figure 2). There was some use of other indirect strategies, namely, disagreement, statement of principle, advice, avoidance, change of topic, wish, and sarcasm. However, due to the relatively low frequency of these strategies, they are not discussed in depth.

Figure 2 Frequency of indirect refusal strategies
The Korean participants used the strategies of reason and regret/apology (18.5% and 17.0% of all refusals, respectively) more than the Norwegian participants (15.3% and 13.6%, respectively). In fact, Korean refusals often consisted of strings of several reasons in addition to other strategies, and reasons and regrets/apologies often co-occurred, as illustrated in (6), where the participant is refusing a friend's invitation to a birthday party: The Norwegian participants used over twice as many hedges (e.g., unfortunately or sadly) as did the Korean participants. Alternatives, however, were used with a similar frequency and in a similar way by both groups of participants and included suggestions of another option or time. A typical use of an alternative is illustrated in (7) Plain indirect strategies were used by the Korean participants to a greater extent (4.9% of all strategies) than by the Norwegian participants (3.2%). Nevertheless, when using plain indirect strategies, both groups tended to mitigate them by additionally employing a regret/apology or a reason, as exemplified in (8)

Adjunct strategies
Adjunct strategies were used extensively by both groups. They comprised 35.2% of all strategies used by the Norwegian participants and 35.8% of all strategies used by the Korean participants. The most frequently employed adjuncts were positive opinion and gratitude, whereas willingness and other strategies (solidarity/empathy and agreement) were used to a lesser extent. Figure 3 summarizes the use of adjunct strategies by the two groups.

Figure 3 Frequency of adjunct refusal strategies
Norwegians and Koreans displayed very similar patterns of use of the positive opinion strategy. In the peer scenario, positive opinion statements were typically related to the birthday party and birthday wishes, as illustrated in (9) In the professor scenario, positive opinions were usually focused on some aspect of the student's own performance or on good experiences associated with the course taken from the requesting professor. An example is provided in (10) Example 10 also provides a good illustration of a typical use of the strategy of gratitude, which, in the professor scenario, often served both as an opener and a closure to the refusal. In the peer scenario, however, gratitude was more commonly expressed at the beginning of the response to thank for the invitation to the party, as in (11) The last adjunct strategy type to be discussed here, willingness, was used by the Korean participants nearly twice as frequently (4.8% of all strategies) as by the Norwegian participants (2.5%). Typically, this strategy followed the pattern "I would . . ., but . . ." and was accompanied by a specific reason. An example is provided in (12)

Discussion
This study compared how multilingual Korean (N = 81) and Norwegian (N = 62) learners of English construct refusals. Although there were statistically significant differences in the frequency of direct, indirect, and adjunct strategies used by the two groups (research question 1), the effect sizes were small. Statistical differences were also found in the average number of strategies used in an equal and unequal social status situation (research question 2); however, the effect sizes were once again small with the exception of indirect strategy use in equal status scenarios. In such scenarios, Koreans used more indirect strategies than the Norwegian participants, and the difference had a small to medium effect size (d = 0.62). Finally, there were more similarities than differences in the way the direct, indirect, and adjunct strategies were used (research question 3). The consistently small effect sizes indicate that the differences between the two groups were not of great magnitude (despite statistical significance), which gives support to the argument that multilingual speakers' pragmatic choices are more complex than predictions of refusal strategy choices based on their L1s may suggest. Below, we provide a more detailed summary and discussion of the main findings.
Research question 1 focused on the extent to which the frequency of direct, indirect, and adjunct strategies differs between Korean and Norwegian users of English. Independent samples t-tests indicated significant mean differences in the use of direct and indirect strategies by the two groups. Similar to previous research, the Korean participants showed preference for indirect strategies (Jung & Kim 2008;Kwon, 2004;Lyuh, 1994). The Norwegian participants, on the other hand, tended to choose more direct strategies. Surprisingly, however, no statistically significant differences were found in the average number of strategies per refusal, nor in the use of adjunct strategies. This finding is at odds with the prediction grounded in Hofstede's (2001) assumption related to status consciousness, namely that Koreans can be expected to use more strategies than Norwegians to soften their refusals. Factors other than cultural differences between the two groups, which we attempt to identify below, may offer a possible explanation with regard to the lack of significant difference in the frequency of the refusal strategy choice.
Research question 2 set out to examine the extent to which Korean and Norwegian users of English employ different strategies in an equal and unequal social status situation. In the present study, the Korean participants employed significantly more indirect strategies both in the equal and the higher status scenario, suggesting that maintaining face was equally important in both situations. Norwegians, on the other hand, tended to use more direct strategies in response to both scenarios, which may be a result of strategy transfer from the first language (Brubaek, 2012), and a manifestation of their egalitarian views and an expectation that the person being refused should respect their right to freely express personal opinions (Hofstede, 2001).
Research question 3 asked about the differences in the types of strategies employed by the two groups. Overall, the two groups also displayed several similarities in the types of direct and indirect strategies they employed. For instance, both groups utilized negation of proposition to a similar extent and tended to employ alternatives in a similar way. However, some differences were also found. Concerning direct and indirect strategy types, the Korean participants used more reasons in comparison with the Norwegian participants, which is consistent with the expectations based on previous studies (e.g., Kwon, 2004;Lyuh, 1994). On the other hand, the Norwegian participants were the only ones to use bluntness, a direct strategy which indicates that free expression of opinions is accepted by this group. These findings are consistent with the assumption that, due to the Korean culture's low score on the cultural dimension of individualism and high score on power distance (Hofstede, 2001), Koreans may work harder to avoid the loss of face in potentially face-threatening situations. There was also a difference in the use of hedging, but, contrary to the expectations, it was the Norwegian participants who used hedges more. This contradicts the prediction that the Korean participants would employ longer sequences of refusal strategies and more mitigating moves. Therefore, it is difficult to determine the extent to which cultural differences affect the observed variation between the groups. It is possible that because all participants were multilinguals with the majority of them having spent some time in the TL culture, their cultural norms of politeness have changed; they have developed multilingual pragmatic competence and thus become more flexible in their choice of pragmatic strategies (Cenoz & Hoffman, 2003;Kecskés, 2015;Safont-Jordà, 2005). Their performance on the study tasks was likely affected by complex factors such as their attitudes to English and exposure to and competence in other languages in addition to their native language cultural norms. It is also impossible to dismiss the impact of fluctuation of cultural values over time and individual differences in pragmatic norms as potential variables that influenced our findings.
Overall, the findings suggest that while native cultural and linguistic backgrounds may have some impact on the production of face-threatening acts such as refusals, multilinguals' pragmatic performance may also diverge from that of monolingual speakers. This can be attributed to the influence of other languages in their linguistic repertoires and the socio-cultural requirements and behavior patterns of those languages (Kecskés, 2015), as well as other factors such as the order of acquisition, extent and mode of exposure, language learning goals, and attitudes towards the L3 (Stavans & Webman Shafran, 2018).
It is important to acknowledge that the present study had several limitations that have to be considered when interpreting the findings. The most important weakness pertains to the use of DCTs as the exclusive means of data collection. While DCTs allow expedient collection of large samples of speech acts that may be otherwise underrepresented in naturalistic data, the responses may diverge from the refusals the participants would have produced in natural settings.
Future research should combine DCTs with other measures, such as role-plays, in order to triangulate the data. Moreover, we did not undertake a qualitative analysis of the content of the refusals of the two groups in this paper. Such an analysis could reveal further similarities and differences in refusal performance, and could, for instance, examine whether reasons given in support of refusals referred to external or personal factors. Another limitation is the age group and the social background of the respondents. As they were all college students, their responses are not entirely representative of all Norwegian and Korean learners of English. Additionally, the majority of the Norwegian respondents were females, and it is possible that gender plays a complex role in the performance of refusals, in particular when confounded with factors such as social distance. Therefore, future studies should include groups of different ages and a better gender balance. Finally, the participant sample of this study was diverse in terms of the length of residence in an English-speaking country and self-reported ability to communicate in English, both of which may have caused some amount of variation within the groups.

Conclusion
The aim of the present study was to investigate differences and similarities in pragmatic performance by Korean and Norwegian multilingual speakers of English with a focus on refusals. As English has more non-native than native speakers, and as we oppose the use of the monolingual norm as a yardstick to measure non-native performance, we purposely selected participants who were multilinguals, that is, those who spoke another language in addition to their native language and English. Two major findings were reported. First, the groups differed significantly in the frequency of direct and indirect strategy use, but the differences in the average number of strategies per refusal and the frequency of adjunct strategies were not statistically significant. Secondly, the analysis of the different types of direct, indirect, and adjunct strategies revealed both similarities and differences between the two groups. At the outset, it was hypothesized that due to differences on the cultural dimensions of power distance and individualism, the two groups could be expected to produce refusals that differ in the level of directness and the amount of mitigation. This hypothesis was only partly confirmed by the findings, as both similarities and differences in the performance of refusals were identified.
Taken together, these findings suggest that multilinguals' pragmatic performance is a complex phenomenon affected by a range of factors, and the first language cultural and pragmatic norms cannot be taken as a single predictor of performance in other languages they know. Rather, it is possible that the pragmatic norms they apply blend into a unique system that also fluctuates depending on the language being used, the context of the interaction, and the intentions of the speaker. To quote Kecskés (2006), multilinguals develop "a pragmatic style that does not exactly match that of either language. Rather, this style is a unique synergic mode of communicative behavior that is the result of conceptual blending" (p. 374).