Pragmatic Markers Produced by Multilingual Speakers : Evidence From a CLIL Context

The purpose of this study is to investigate the production of pragmatic markers (PMs) by multilingual students in a CLIL context. Previous studies have analyzed pragmatic competence in multilingual settings (e.g., Cenoz, 2003; Martín-Laguna & Alcón-Soler, 2015; Portolés, 2015; Safont & Portolés, 2016). However, to the best of our knowledge, no previous study has investigated the use of PMs across languages at the oral level in the multilingual classroom. As suggested by Nashaat-Sobhy (2017, p. 69), there is a need for studies that support or refute whether CLIL helps students communicate more appropriately. In an attempt to fill this gap, the overall aim of this research study is to explore how multilingual students use PMs across languages –namely Spanish, Catalan and Englishin terms of frequency and type of PM. Participants were 19 Spanish students in an instructional context where three languages are in contact, namely English, Catalan and Spanish. They completed a language background questionnaire and comparable oral decision-making tasks carried out in pairs, one task in each of the target languages. The analysis explored the frequency and type of PMs (i.e. textual and interpersonal markers). Findings revealed significant differences in the frequency and type of both interpersonal and textual PMs across the three languages, shedding some light on the particular characteristics of the pragmatic competence of multilingual learners in a CLIL setting.


Introduction
The promotion of linguistic diversity in education has always been an important consideration in implementing Content and Language-Integrated (CLIL) programs.The CLIL approach, in which an additional language is used for the learning and teaching of both content and language, is believed to improve language competence in the target language by focusing on meaning and communication in the classroom (Lasagabaster, 2008).The motivation behind the spread of CLIL in Europe is due to two main factors.The first one is the political mandate in the European Union, which encouraged all citizens to learn two community languages in addition to their mother tongue (European Commission, 1995).The second one is the impact of globalization on education, which has resulted in a growing interest in international mobility and language learning programs.In Spain, educational authorities have embraced this dual approach in an attempt to enhance the acquisition of English as a foreign language (EFL) and to promote multilingual practices (see Juan-Garau & Salazar-Noguera, 2015, for a discussion of the implementation of CLIL in multilingual educational settings in Spain).In this context, different studies have explored second language (L2) acquisition by multilingual learners, revealing their advantage over monolingual students for the acquisition of different linguistic abilities such as requestive behavior (Cenoz, 2013).Nevertheless, there is scarce empirical evidence on the acquisition of L2 pragmatic competence by multilingual students in CLIL programs.To address this concern, the present study explores the use of PMs across languages-namely Catalan, Spanish and English-by multilingual students in a CLIL program in Spain.The production of pragmatic markers in speech is used to measure pragmatic competence because they " […] contribute to the pragmatic meaning of utterances and thus play an important role in the pragmatic competence of the speaker" (Müller, 2005, p. 1).

CLIL and pragmatic competence
Although the area of interlanguage pragmatics (ILP) in CLIL contexts is still under-researched, a few studies have examined the CLIL setting as a potential environment for the enhancement of foreign language (FL) learners ' pragmatic competence (e.g., Dalton-Puffer, 2005;Dalton-Puffer & Nikula, 2006;Nashaat-Shoby, 2018;Nikula, 2008).Dalton-Puffer and Nikula (2006) observed that direct request strategies were used regularly in the classroom and that students relied on direct requests when asking for information and indirect for action.In a later study, Nikula (2008) examined secondary-school learners' classroom interaction in Finnish CLIL Physics lessons.The main focus of the study was on face-threatening acts like disagreements and misunderstandings with the teachers.Results showed that students used hesitations, disagreements with yeah but-formulations and preparatory conditions with grounders when requesting (e.g.Can I ask something?).In a more recent study, Nashaat-Shoby (2018) compared the acquisition of request modification in CLIL and non-CLIL secondary school students.By means of a written discourse completion test (WDCT), the researcher analyzed learners' production of requestive pragmatic moves according to their pragmatic functions i.e. softeners and aggravators.Results revealed significant statistical differences between both groups by indicating that the CLIL group had a wider repertoire of request modification strategies.Nashaat-Shoby (2017) reported similar results.The researcher compared how learners in CLIL, non-CLIL and EFL classes modified their requests by means of a WDCT.Results showed that CLIL students used more softening and aggravating strategies.However, the comparison between CLIL, non-CLIL and older EFL students with extra hours of English instruction showed that CLIL was not the decisive factor, but rather the students' continued exposure to English over time.In a similar vein, Nikula (2002) concluded that CLIL students did not use indirect pragmatic resources when interacting in the classroom.Taking into account the above mentioned studies, there is no certainty that the CLIL context benefits learners' pragmatic learning.In addition, there is a need to investigate CLIL students' pragmatic performance beyond the domain of speech acts.The present study addresses this concern by exploring the use of PMs by multilingual learners across languages in the CLIL setting.

Multilingual Students' Pragmatic Competence
Previous studies have explored L2 acquisition by multilingual learners (e.g., De Angelis & Dewaele, 2011; Gabrys-Barker, 2012).However, few studies have analyzed pragmatic competence in multilingual settings (exceptions include Cenoz, 2003;Martin-Laguna & Alcón-Soler, 2015;Portolés, 2015;Safont & Portolés, 2016;Safont, 2005;Safont & Alcón-Soler, 2012).Results on pragmatic competence in multilingual settings suggest that multilingual learners have an advantage over monolinguals in producing requests (Cenoz, 2003;Safont, 2005), request modifiers (Safont & Alcón-Soler, 2012) and refusals (Alcón-Soler 2012), and in using translanguaging practices (Safont & Portolés, 2016).Multilingual students' pragmatic competence has also been studied in terms of awareness and comprehension of target pragmatic features.For instance, Portolés (2015) analyzed pragmatic awareness and comprehension of requests in infant and primary education by means of an audiovisual pragmatic test.Learners were asked to report the appropriateness of request moves in Spanish, Catalan and English.The findings revealed that learners were able to judge appropriate requests moves in the three languages and that awareness was influenced by learners' multilingual proficiency.

Pragmatic Markers
The present study focuses on one specific aspect of multilingual students' pragmatic competence; that is, pragmatic markers (PMs).PMs such as okay, yeah, you know, I mean or well have been broadly defined as short linguistic elements that do not convey particular semantic meanings but which carry out significant pragmatic functions in conversation (e.g.Aijmer, 2013;Andersen, 2001;House, 2003;Romero-Trillo, 2012).Different scholars have acknowledged the role of PMs as key elements of communicative competence (e.g.Romero-Trillo, 2012, 2018).Nevertheless, they still represent an underresearched area in SLA.In L2 classroom research, existent findings point out that appropriate use of PMs in English enhances oral fluency (Hasselgreen, 2004;Shively, 2015).Indeed, the use of PMs is associated with learners' proficiency level; more proficient L2 students seem to use a higher ratio of PMs and with a wider variety of functions (Neary-Sundquist, 2014).Nevertheless, when compared with native-speakers (NSs), EFL learners tend to use a lower ratio of PMs, and with different functions (Müller, 2005;Romero-Trillo, 2002).
A key aspect of PMs is their multifunctionality; that is, one PM may serve different functions according to the context in which it is uttered.Because of this, different functional approaches to PMs have been proposed.Some view PMs as serving multiple functions, from speech management to politeness (Fischer, 2006), others argue that PMs have either discursive or interpersonal functions (Ament, Pérez-Vidal & Barón, 2018;Andersen, 2001;Brinton, 1996), while numerous researchers only distinguish the interpersonal nature of PMs, differentiating between PMs and discourse markers (Romero-Trillo, 2012).The present study adopts the approach to PMs as performing two main functions: textual and interpersonal.On the one hand, textual PMs contribute to the organization of the discourse, facilitating the interpretation to the interlocutor (e.g., however, anyway).On the other hand, interpersonal PMs manage the relationship between the speaker and the message (e.g.definitely, I think) and between the speaker and the hearer (e.g., listen, that's right).Only a few studies have compared the use of textual and interpersonal PMs by FL students, revealing higher amounts of textual PMs than of interpersonal ones (Ament et al., 2018;Fung & Carter, 2007), probably because of the fact that textual markers are more present in the FL classroom instructional input.
Regarding the multilingual classroom context, to the best of our knowledge, no previous study has investigated the use of PMs across languages at the oral level, a gap that the present study attempts to cover.At the written level, Martín-Laguna and Alcón-Soler (2015) focused on the use of textual PMs in argumentative essays written in Catalan, Spanish and English, revealing no significant differences in the ratio and variety of textual PMs used in the three languages.Nonetheless, the question then still remains as to whether there are differences in the use of PMs across languages in oral communication by multilingual learners.

Research Question
The present study sheds new light on the acquisition of L2 pragmatic competence by multilingual speakers by exploring the use of PMs across languages -namely Catalan, Spanish and English -by multilingual students in a CLIL program in Spain.To address this purpose, the following research question was formulated: • Are there differences across languages (in this case, Catalan, Spanish and English) in the use of PMs by multilingual students, in terms of frequency and type of PM?

Method
In this section, the methodology of the study is described, starting with the participants, the data collection instruments, the procedure, and finally the analysis.

Participants
Data were collected from one state secondary school in Castelló (Spain) participating in a CLIL multilingual program.A total of 19 adolescent learners of English, 15 to 16 years old, participated in the study.Results from the language background questionnaire revealed that their first language (L1) was predominantly Spanish (n = 10), followed by Spanish and Catalan (n = 7) and a minority of speakers of Romanian (n = 2).As for the participants' language use, they all were Spanish/Catalan bilinguals, while English was reported as a third language (L3).Regarding their previous L2 experience, learners reported having received at least 10 years of formal instruction in English since their first year in Primary education.They had just finished grade 10 and participated in a CLIL multilingual program (Spanish/Catalan/English) over the academic year 2017/2018.In relation to the number of subjects taken in English, all of them received tuition in History (95 hours) and Technology (110 hours) over the academic year.Besides History and Technology, students received Catalan tuition in eight subjects (approximately 760 hours) and Spanish tuition in two subjects (approximately 190 hours).All in all, their English proficiency level was in average intermediate (Note 1), as measured by their institutional evaluation system.

Data Collection Instruments
A language background questionnaire was used to gather information on participants' linguistic background.This questionnaire asked for demographic information, with a focus on participants' L1 and self-reported English proficiency.
To explore participants' use of PMs across languages, an oral decision-making task was designed.In this task participants engaged in conversation with another participant, requiring them to reach an agreement through negotiations to decide on the level of agreement with a specific statement (1 = strongly agree, 2 = agree, 3 = disagree, 4 = totally disagree).The task was performed in pairs three times, once in Spanish, once in Catalan and once in English.The structure of the task, task type and topic were kept constant in order not to influence PM use.The topics in the three languages concerned different aspects related to a common topic, i.e. social networks (see Table 1).The tasks were performed first in Catalan, then in Spanish and, finally, in English.

Procedure
Data collection took place at the end of the 2017/2018 academic year.For the present study, the data come from an oral decision-making task and a language background questionnaire.The instruments were administered in paper format during face-to-face sessions.Participants were informed that their responses would remain anonymous and that they had to be honest while discussing the topic as naturally as possible with their partner.
All the participants were given ten minutes for the language background questionnaire.After completion, participants were asked to work in pairs in order to perform the oral tasks, resulting in pair-conversations and a conversation among three participants, which were audio recorded.The instructions were provided by one of the main researchers.Participants had fifteen minutes for the oral decision-making task, i.e. five minutes to discuss each language prompt (see Table 1).In this time frame, they had to express their interest in the statements and reach a final agreement by responding on the Likert scale.

Analysis
The conversations were transcribed manually, and revised by the two main researchers to ensure accuracy.Three corpora were created, one with students' interactions in Catalan, one in Spanish, and one in English.Next, coding of the PMs was performed with the assistance of the software F4 Analyse.A functional approach to coding PMs was followed, taking into account the multifunctionality of PMs; that is, the fact that one same graphic PM may have different functions according to the context.We categorized each PM as either textual or interpersonal following the categorization proposed by Ament et al. (2018), and included two additional functions of interpersonal PMs found in our corpora: fillers and general extenders.On the one hand, fillers included I don't know, like and it's like in English, pues, es que, como, en plan, no sé, and ya in Spanish, and pues, no sé, es que and vull dir in Catalan.On the other hand, hesitators included I think, ehh, and uhmm in English, ehh, ahmm, no sé, pues, and es que in Spanish, and ehh, uhm, no sé, and es que in Catalan.Additionally, we created the category "transfer PMs", which included both textual and interpersonal PMs in a different language than the one being used.Coding was validated via inter-rater reliability, the raters being experts in L2 pragmatics After a first data coding, a recruited scholar independently coded 28% of the data, the agreement rate resulting in 88%.
After coding the PMs, ratios of use of PMs by each participant in each language was calculated in relation to the total amount of words uttered (2,632 words in English, 3,240 words in Spanish, and 2,354 words in Catalan).In doing so, we obtained percentages of frequency of PM appearance.The resulting data were normally distributed, as indicated via a series of Shapiro-Wilk's tests (p > .05),allowing the performance of parametric tests.We used three-way analysis of variance (ANOVA) and post-hoc Tuckey multiple-comparison tests to determine whether multilingual students in a CLIL setting showed differences in their use of PMs across the three languages, that is, Catalan, Spanish and English.Results were considered significant at a value of p < .05.

Results and Discussion
The framing research question of the study asked how multilingual students use PMs across languages (in this case, Catalan, Spanish and English), in terms of frequency and type of PM.A first glance at Table 2 shows a higher average of PMs used in English, followed by Catalan, and a least amount of PMs used in Spanish.To statistically test the validity of such assumption, a three-way ANOVA test was conducted, indeed revealing significant differences in the use of PMs between at least two of the target languages [F(2, 54) = 14.073; p = .00].More specifically, a post-hoc Tuckey multiple-comparison test pointed out differences between English and Catalan (M difference = 6.63; p = .003),and between English and Spanish (M difference = 9.84; p = .000),but not between Catalan and Spanish.Indeed, these findings reveal that multilingual EFL students use substantially more PMs in English than in their L1s.Different hypotheses could be drawn from this result.On the one hand, a high rate of PMs in learners' interlanguage could be attributed to a high presence of these in the instructional input, since previous studies have pointed out the effects of instruction on PM use in a multilingual setting (Martín-Laguna & Alcón-Soler, 2018;Neary-Sundquist, 2014).On the other hand, learners may have higher pragmatic awareness in English than in the other languages, which makes them particularly notice those PMs and incorporate them in their speech (Neary-Sundquist, 2014).
A closer look at PM use across languages is illustrated in Table 3, which displays descriptive statistics of textual and interpersonal pragmatic markers.As for interpersonal PMs, the ratio of use was significantly higher in English than in the other two languages (M difference with Spanish = 5.08; p = .000;M difference with Catalan = 3.28; p = .025).These results could be explained by the exposure to textual and interpersonal PMs the multilingual students have.On the one hand, textual PMs are more present in instructional input (Martín-Laguna & Alcón-Soler, 2015).Given that the multilingual program in which students were immersed included Spanish and English, the limited instruction in Catalan may have limited their use of textual PMs in such language.On the other hand, students may have acquired interpersonal PMs in English either through exposure to out-of-school settings, or thanks to the CLIL program in which they are immersed.Unlike non-CLIL instruction, where input in the L2 tends to be limited to EFL classes, CLIL students are exposed to more language registers (Lorenzo, Casal & Moore, 2009) since they are encouraged to communicate in an additional language with their peers and with the teacher.
When compared, the use of textual and interpersonal PMs was not significant, as indicated by a series of 2-tailed Pearson correlation tests.It was not significant either for overall use of textual and interpersonal markers (p = .930),nor in each of the languages -Catalan (p = .106),Spanish (p = .143)and English (p = .954).This finding provides support for the perception of textual and interpersonal PMs as different categories at the oral level.They represent different communication strategies which should be considered independently, as has been done by numerous scholars (Ament et al., 2018;Fung & Carter, 2007).
Next, we explored differences in the use of PMs across languages, in terms of the different subcategories of textual and interpersonal pragmatic markers.Table 4 illustrates the frequency of use of each type of PM in the three languages, indicated both with ratio of frequency and percentage of frequency with respect to the total words included in the corresponding corpora.A series of ANOVA tests indicated that there were significant differences across languages in multilingual learners' use of fillers [F (2,54) = 6.680; p = .003and hesitators [F (2,54) = 12.745; p = .000].Nevertheless, the analysis did not reveal significant differences in the rest of textual and interpersonal subcategories.Regarding fillers, a post-hoc Tuckey test revealed that there were differences between Spanish and the other two languages (M difference with English = 1.023; p = .003;M difference with Catalan = .770;p = .029),but not between Catalan and English.This finding may indicate that the use of fillers is a communication strategy students use when they share the same L1 with their interlocutor, inserting PMs that both of them are knowledgeable of, and therefore shortening the relationship between both.As for the use of hesitators, a post-hoc test indicated that there were differences between English and the other two languages (M difference with Catalan = 2.33 p = .015;M difference with Spanish = 4.07; p = .000),but not between Catalan and Spanish.By looking at Table 4, one can observe that learners use a higher number of hesitators in English, a communication strategy multilingual learners seem to use in the additional language more than in their L1s.This, indeed, may be explained by their lack of proficiency in English as an additional language.
In addition to textual and pragmatic markers, our corpora included the category of transfer PMs; that is, PMs that are not in the language being used.In both the English corpus and the Catalan one, transfer PMs were in Spanish, while in the Spanish corpus there were no instances of transfer PMs.Regarding the English corpus, from the total amount of transfer PMs (n = 8), 87.5% were textual (n = 1), and 12.5% interpersonal, while the Catalan corpus (transfer PMs n = 43), 55.8% were textual and 44.2% interpersonal.As can be inferred, a Fisher's exact test revealed significant differences between the use of transfer PMs in English and in Catalan [χ(1) = 58.615,p = .00].This finding sheds some light on the particular characteristics the multilingual learner has for the acquisition of different linguistic abilities (Pauwels, 2014).
In summary, findings from the present study indicated that multilingual students use PMs differently across languages at the oral level.These differences are in terms of both textual and interpersonal pragmatic markers, and are especially significant in the use of hesitators and fillers.

Conclusions
In conclusion, the present study has revealed differences across languages in the use of PMs by multilingual learners, in terms of both frequency and type of PM.More particularly, the results have shown that students use more PMs in their L3, English in this case, than in their mother languages.Additionally, differences across languages were observed in the use of textual and interpersonal PMs.Regarding textual PMs, which support the speaker in organizing the discourse, students used significantly less PMs in Catalan.As for interpersonal PMs, which relate the speaker with the proposition and with the interlocutor, our participants used a significant higher rate of interpersonal PMs in English, as compared with the other two languages.
The reported findings have pointed out the difference between use of PMs in written and in oral form by multilingual students.At the written level, Martín-Laguna and Alcón-Soler (2015) pointed out that students use PMs in L2 English, L1 Catalan and L1 Spanish in the same way (i.e. with a similar frequency and with similar functions).In contrast, the present study has revealed significant differences across such languages at the oral level.Focusing the same instructional context, i.e. a secondary school in the Valencian community in Spain, this investigation showed that students used a higher amount of PMs in English, as compared with their L1s.
Taking into account such differences, we hypothesize that multilingual students may use PMs to compensate for lack of oral proficiency, filling in their discourse with linguistic particles with little semantic meaning but which serve them to carry out discourse-organization and interpersonal functions.
This study presents, however, some limitations.The most obvious limitation is the number of participants, which prevents a clear generalized statement about the use of PMs by multilingual learners in a CLIL setting.In relation to the instrument administered to explore learners' use of PMs across languages, an oral decision-making task was used in order to promote oral interaction.Although this instrument involves real-time conversations, we acknowledge that oral decision-making tasks may limit authenticity of interaction.Participants were given specific prompts to discuss and were influenced by time pressure, which may have affected the use of pragmatic markers.Nevertheless, oral decision-making tasks represent a good option to collect data on learners' oral production of pragmalinguistic features, as was the case in this study.Bearing all the above in mind, the results of this study invite researchers to conduct more studies on PMs in other multilingual settings.For instance, it would be interesting to do follow-up studies and explore factors that may determine PM use, such as instruction (Martín-Laguna & Alcón-Soler, 2018;Neary-Sundquist, 2014), teachers' proficiency level (Dalton-Puffer, 2005, 2007;Nikula, 2002), and attitudes toward languages (Portolés, 2015).Finally, research should also investigate differences in CLIL and non-CLIL students, since studies have revealed the advantage of CLIL over non-CLIL settings (Menezes & Juan-Garau, 2014;Jiménez-Catalan, Ruiz de Zarobe & Cenoz, 2006).

Table 1 .
Topics used in the oral decision-making task Table 2 displays ratio scores and standard deviations of total amount of PMs used in each language.

Table 3 .
Descriptive statistics of textual and interpersonal PMs used across languages A series of ANOVA tests revealed significant differences among at least two of the groups in the use of both textual [F (2,54) = 6.009 ; p = .004],and interpersonal PMs [F (2,54) = 8.949 ; p = .000].Regarding textual PMs, a post-hoc Tuckey test indicated that multilingual students significantly used less PMs in Catalan than in the other two languages (M difference with Spanish = -2.32;p = .014;M difference with English = -2.45;p = .009).

Table 4 .
Descriptive statistics of subcategories of textual and interpersonal PMs