The Development of Pragmatic Markers in English as a Second Language: Do Age and Learning Context Matter?

Abstract

The present study explores the development of pragmatic marker (PM) use by children and adult L2 English learners in two learning contexts: study abroad (SA) and at home (AH). The study involved a group of 35 Catalan/Spanish girls (aged 11 to 13) learning English AH (n = 16) and abroad in Ireland (n = 19), and a group of 16 adult students aged 19–31 learning English in the UK and Ireland (n = 10) and at their home university in Barcelona (n = 6). To test their pragmatic development, the use of PMs was prompted through pre-test and post-test semi-structured interviews. The results indicated an effect of both age and context on PM development. Children in the SA context increased their use of some PMs, whereas their peers who remained at home did not show any development. Regarding the adults, both SA and AH participants increased their use of specific PMs. These findings shed some light on a topic that has not received scholarly attention in the field of L2 pragmatics: the development of PM use by children. Additionally, they contribute to the very small body of longitudinal studies on the development of PM use during SA.

1. Introduction

Nowadays, studying abroad is part of the academic trajectories of many university students due to globalization and the internationalization of higher education. However, young learners in some schools also have the opportunity to engage in similar experiences. Research exploring the effects of ‘study abroad’ (SA) experiences on L2 development has usually found that learners tend to improve their L2 competence after the SA period (see Llanes 2020 for a review). Similar findings have been reported for L2 pragmatic learning (e.g., Barron 2019; Pérez-Vidal and Shively 2019). However, it has been claimed that when it comes to pragmatic development, not only the amount of exposure seems to be an important factor to consider, but also the pragmatic forms to be learned, the opportunity to interact with learners with a variety of social distance and power, the contextual situations they find themselves immersed in, or even their own individual differences (Taguchi 2015; Sánchez-Hernández and Alcón-Soler 2019).

Most of the research which has examined SA and L2 pragmatics has mainly focused on young adults and adults (e.g., Alcón-Soler 2015; Bardovi-Harlig and Bastos 2011; Barron 2019; Félix-Brasdefer and Hasler-Barker 2015). Very few studies have analyzed such a combination in young learners (e.g., Achiba 2003; Ellis 1992), and to the best of our knowledge, no studies have investigated the effects of SA experiences in L2 learning in both young learners and adults in the same study. This seems relevant to explore, since as pointed out in Bialystok’s Two-dimensional model (1994), adult learners might have an advantage when learning L2 pragmatics, since they have already developed their first language (L1) social and cultural representations, thus they “only” need to learn such representations in the L2. However, in the case of children, they still have to build those representations in their L1 at the same time that they are acquiring the L2. Thus, it is necessary to explore whether the SA experience affects their L2 pragmatic development at different ages.

As regards the pragmatic moves analyzed, speech acts have been extensively investigated (Li et al. 2022), as in many studies in interlanguage pragmatics (ILP), leaving other pragmatic aspects generally unexplored. This is the reason why the present study focuses on pragmatic markers (PMs), which have been mainly examined among adult learners (e.g., Magliacane and Howard 2019). The rationale for having PMs as the focus of investigation is that they are frequently used by native speakers (NS), thus learners going abroad might find themselves being exposed to them, as well as in need to use them. However, they have also been considered a difficult pragmatic feature to be acquired, since they tend to be culturally marked in that in order to use them appropriately, learners need to know the specific linguistic features to be performed in a particular context to avoid pragmatic failure (Ament et al. 2020). Therefore, the aim of the present study is to explore to what extent Catalan/Spanish children and adults acquire English PMs during an SA program in Ireland and the UK.

The paper first provides an overview of previous studies on L2 pragmatic learning across different educational contexts and at different ages, and then it focuses on the acquisition of PMs in SA and AH contexts. In Section 3, the study conducted is presented (with information about the participants, instruments, and measures). Section 4 reports the results found in the data from the current study, and finally Section 5 and Section 6 follow with a discussion of the findings and the conclusions that can be drawn from this paper.

2. Literature Review

2.1. L2 Pragmatic Learning across Contexts and Age

Many activities that we do nowadays require us to be multilingual competent speakers, as most of them involve interactions with both native and non-native speakers (NSs/NNSs) (Sánchez-Hernández and Alcón-Soler 2019). As claimed in the field of ILP, being linguistically competent alone does not guarantee competence in a second or foreign language (L2); it is the combination of linguistic competence—that is, mastery of the lexicon, syntax, semantics and phonetics of a language (CEFR 2001)—with pragmatic competence—i.e., understanding “how-to-say-what-to-whom-and-when” (Bardovi-Harlig 2013, p. 68)—that leads to successful interaction with others (Llanes 2020). However, being pragmatically competent is not an easy task, as it requires both pragmalinguistic competence (linguistic knowledge used to convey pragmatic meaning) and sociopragmatic competence (sociocultural knowledge about when, where, and how to use such linguistic knowledge) (Félix-Brasdefer 2021).

In English-as-a-foreign-language (EFL) contexts, research has shown that instruction in L2 pragmatics is beneficial for developing pragmatic competence (Bardovi-Harlig 2020), especially since exposure and opportunities for interaction in the L2 are limited compared to other contexts such as immersion settings. Research in ILP has particularly focused on SA experiences, where students of different ages and proficiency levels immerse themselves in a community (for short or long periods of time) where the target language (TL) is used. Generally, studies of this nature have reported positive effects of SA experiences on L2 pragmatic learning (e.g., Barron 2019; Pérez-Vidal and Shively 2019 for a detailed review), attributed to factors such as exposure to local pragmatic norms and engagement in real-life interaction practices (Taguchi 2018).

Over the past two decades, there has been increasing interest in examining the effects of SA on pragmatic learning, especially among undergraduate students. These studies have focused on both receptive and productive pragmatic skills and have included a wide variety of speech acts. For instance, Taguchi (2008) examined the development of comprehension of conventional and non-conventional implicatures in 64 Japanese college learners of English divided into three groups: two staying at home (AH) and one with a SA experience of one year. The results revealed positive effects of SA experiences on the acquisition of non-conventionalized implicatures and routine expressions. Another study by Alcón-Soler and Sánchez-Hernández (2017) on pragmatic routines with 122 English students who spent a semester in the US similarly found that the SA experience increased learners’ L2 pragmatic knowledge and recognition of pragmatic moves. Regarding the development of productive pragmatic skills, Ren (2015) studied 40 L2 Chinese learners with varying lengths of stay in the target community, finding that longer stays sometimes led learners to deviate from native speaker norms due to potential lack of sociopragmatic knowledge or insecurity. Additionally, Li et al. (2022) studied American students of different proficiency levels enrolled in a SA program in China, observing development in a variety of speech acts during the stay, with proficiency levels also influencing the extent of development.

Regarding children, very few studies have focused on this under-represented population. In this line, Achiba (2003) explored how the use of requests developed in a 7-year-old Japanese child after a 17-month stay in Australia. The development went through four main phases. In the first phase, she used formulaic sequences through routines and patterns. In the second phase, more complex patterns were used, and more indirect requests also started to emerge. The third phase was characterized by a wider variety of modal verbs for requests addressed to speakers with different ranks. In the last phase, more indirect requests appeared along with the use of mitigating devices. Another study by Jones (2007) explored the development of interactional markers, politeness, and address forms in three L2 Japanese learners (L1 English) aged seven, five, and two over a year. The findings were compared with those from adult learners, revealing that the learning process was faster in the case of children than in adults. At a time when the study of L2 pragmatic development in young learners is increasing (e.g., Myrset 2021; Savić et al. 2021), it seems relevant to further investigate how they develop pragmatic competence in SA contexts.

2.2. Pragmatic Markers in Study Abroad and at-Home Contexts

One area within the study of L2 pragmatic development during SA that has received very little scholarly attention is that of PMs. PMs such as well, you know, or I mean, are linguistic constructions recurrent in spoken speech which contribute to the smoothness of interaction (D’Arcy 2017). As different scholars have pointed out, PMs are phonologically reduced and syntactically optional. As such, they can be removed from the discourse without altering the communicative message (Fedriani and Sansò 2017). While PMs do not convey specific semantic meanings, they are used to carry out key communicative functions. They may play textual functions when they contribute to the organization and management of the discourse (e.g., Well used to start a topic of conversation, or Because, So or And used to mark causal relationships). Moreover, they can be used for interpersonal functions when they signal the relationship between the speaker and the message (e.g., Well used to hesitate) or between the speaker and the hearer (e.g., Okay, Right or Yeah used to show that the information has been understood; Great, Exactly or Sure used to show support towards the interlocutor) (Magliacane and Howard 2019; Author 4). Altogether, PMs are a key element in spontaneous speech, as they help speakers plan their speech and contribute to maintaining speech fluency and smooth interaction.

Despite the key role PMs have in oral communication, their use by L2 speakers has not received much scholarly attention (but see Diskin 2017; Diskin-Holdaway 2021; Magliacane 2020; Magliacane and Howard 2019). Studies comparing PM production by L1 users with that of L2 speakers have observed that L1 users employ PMs more frequently and with a wider range of functions. For instance, House (2009) observed that you know, which is commonly employed by L1 English speakers as an interpersonal marker, was used by German speakers mainly for textual purposes. Moreover, L2 speakers tend to use PMs that have a similar equivalent in their L1s. For example, Diskin-Holdaway (2021) showed how her Polish participants used you know more frequently than their Chinese peers, and ascribed it to the presence of a similar PM (wiesz) in Polish.

Research on the effects of SA on the acquisition of PMs has been scarce, but the few existing studies generally agree that exposure to the L2 in a naturalistic context and regular contact with the target-language community enhances PM development (e.g., Beeching 2015; Huang et al. 2023; Liao 2009). The majority of these studies have been cross-sectional and have compared the use of PMs by L2 speakers with that of L1 speakers, as well as the use of PMs by L2 learners with different SA lengths (Diskin 2017; Diskin-Holdaway 2021; Liao 2009; Huang et al. 2023). For instance, Huang et al. (2023) explored the use of the PMs well, like and you know, and investigated the role of proficiency (B2 to C1) and SA experience on the reported PM use. To operationalize SA experience, the study involved a group of SA students in English-speaking countries with SA lengths ranging from 0.2 months to 167.8 months (n = 101), and a group of students with no SA experience (n = 72). The findings pointed to a strong relationship between the SA experience and the frequency of use of the PMs you know and well. These PMs, unlike “like,” are commonly employed for interpersonal functions; this may suggest that an SA experience may especially enhance the ability to use interpersonal PMs.

Longitudinal studies exploring PM development during SA are rather scarce (exceptions include De Cristofano and Badan 2021; Magliacane and Howard 2019; Magliacane 2020), despite the importance of conducting longitudinal research to account for the complex nature of L2 pragmatic development during SA (see Llanes 2020). Magliacane and Howard (2019) and Magliacane (2020), for instance, observed that a period of SA in Ireland resulted in an increase in PM production (of well and like) by Italian users of English, which included SA undergraduate students and au-pair employees. Nevertheless, these participants used PMs predominantly for textual functions (e.g., to buy thinking time), while interpersonal PMs (e.g., reactions to what the speaker said) showed differences compared to a group of Irish speakers. The findings showed a tendency towards target-like PM use aided by contact with L2 users. More particularly, the au-pair group showed a more native-like use of PMs as they lived with Irish families and had extensive exposure to children’s English speech. In contrast, the SA group mainly interacted with other international students. Another example is De Cristofano and Badan (2021), who explored the development of PM use by Belgian–Dutch learners of L2 Italian enrolled in a six-month SA program in Italy. They observed that rather than the frequency and type of PMs, it was the variation in the pragmatic functions of PMs that mostly developed during SA. More particularly, the participants expanded their use of PMs with textual functions. Moreover, they also increased the use of PMs with non-target functions, which pointed out the potential role of L1 transfer.

Regarding at-home contexts, there is a line of research exploring the oral use of PMs in Content and Language-Integrated Learning (CLIL) settings (Herraiz-Martínez and Sánchez-Hernández 2019; Lázaro Ibarrola 2016), as well as in English-Medium Education (EME) contexts (Campillos-Llanos and Gozalo 2014). These contexts, unlike the traditional foreign language classroom, offer additional exposure to English as learners engage with disciplinary content through the use of the L2. As Lázaro Ibarrola (2016) explains, the development of specific linguistic aspects in the L2 seems to be determined by the amount and type of L2 input received by CLIL/EMI students, which is qualitatively different from the input received in a traditional EFL classroom:

“The input provided in CLIL lessons has the aim of conveying knowledge about a subject matter and is, therefore, communicatively more meaningful than the input provided in EFL lessons, which is often artificially manipulated for the sake of teaching the language. In addition, in CLIL lessons the learners become accustomed to using the language for interaction either to ask for contents or forms. […] Accordingly, CLIL students might consider and use the target language as an instrument of communication prioritizing a focus on meaning. By contrast, EFL students might see it merely as a school subject, as an object to be learnt, thus prioritizing a focus on form.”

(Lázaro Ibarrola 2016, pp. 128–29)

Findings on the use of PMs in CLIL/EMI contexts suggest that young learners and adolescents in these settings underutilize PMs and/or use them inappropriately, even at high-proficiency L2 levels (Campillos-Llanos and Gozalo 2014; Hellermann and Vergun 2007; Romero-Trillo 2002). This may be due to a lack of exposure to PMs in the classroom input. For example, Campillos-Llanos and Gozalo (2014) explored the use of PMs by learners of L2 Spanish with different L1s compared to that of L1 Spanish speakers, finding that L2 learners used PMs at a lower rate than L1 users. Nevertheless, Herraiz-Martínez and Sánchez-Hernández (2019) in a study about the use of PMs across languages—namely Catalan, Spanish, and English—by multilingual students in a CLIL program in Spain, found that the students used a higher amount of PMs in L2 English compared with their L1s (Spanish and Catalan). This may be due to the overuse of specific L2 PMs. For instance, the PM “well” appears to be very present in English textbooks and is also frequently used by CLIL teachers (Hellermann and Vergun 2007).

In addition to this, L2 learners in CLIL and EMI contexts seem to code-switch the PMs from their L1; that is, they use L1 PMs while speaking in the L2 (Campillos-Llanos and Gozalo 2014; Hlavac 2006; Lázaro Ibarrola 2016). Lázaro Ibarrola (2016), for instance, compared PM production by a group of Basque-Spanish bilinguals in a CLIL high-school setting with that of their EFL counterparts, finding a similar trend in the use of L1 PMs by the two groups. As the author concludes, the different quality of input CLIL learners received did not seem to make a difference in the production of PMs by young learners.

Altogether, findings on the development of PM use in SA and CLIL/EME contexts point to the beneficial role that exposure to meaningful target-language use and constant contact with L2 users has on the development of PMs. Nevertheless, the majority of the studies mentioned above have focused on adult populations, and little is known about how children acquire PMs while being engaged in SA and CLIL settings. To address this research gap, the present study explores how children and adult EFL learners develop their use of PMs in two learning contexts: study abroad (SA) and at home (AH). The following research questions (RQs) guided this investigation:

• RQ1: To what extent does age (children vs. adults) have an effect on the use of pragmatic markers?
• RQ2: To what extent do age (children vs. adults) and learning context (SA vs. AH) influence the development of PM use?

3. The Study

3.1. Participants

A total of 51 female students participated in the present study, all of them being Catalan/Spanish speakers and learners of L2 English. The sample included 35 young learners and 16 adults.

Table 1. Characteristics and distribution of the participants.

	Age	AH	SA
Children	11–13 (M = 12)	16	19
Adults	19–33 (M = 20.5)	6	10
		22	29

illustrates the age and learning context (SA vs. AH) of these participants.

As we can see in Table 1, the age of the children ranged from 11 to 13 years old (M = 12), and their proficiency level was estimated to be at B1 (according to the Common European Framework of Reference for Languages (CEFR)), as indicated by their classroom placement and performance. They were grouped according to their learning context, with 16 learners in the at-home (AH) setting and 19 in the SA context. These girls were studying in a single-sex private school in the area of Barcelona, Spain. Some of them were given the opportunity to study in Ireland for two months, where they were exposed to naturalistic input both in and outside school. The SA girls were placed in regular Irish schools, which they attended for an average of 25 h per week, with no other Spanish students. Additionally, they lived with Irish families that had children of similar ages. A Catalan teacher accompanied them to solve any problems they might have and to teach them Catalan and Spanish after school so that they did not fall behind in these two subjects when they returned to Barcelona. As for the girls who remained in the AH setting, they received English classes four times per week and attended Science classes in English for two hours per week. Hence, they had a total exposure to English of six hours per week.

The 16 adult participants were all female students enrolled in the third year of the English Studies degree at a public Catalan University in the region of Barcelona. Therefore, they were learning English in an EME setting. Their age ranged from 19 to 33, with an average of 20.5 years old, and their proficiency level was estimated to be C1, as indicated by their classroom placement (they were all enrolled in a C1 English class). Ten of these sixteen participants received an ERASMUS scholarship to participate in a three-month-long SA experience in the UK or Ireland. All of them lived in a dormitory during their time abroad, and they attended classes at their host university for 8 to 12 h per week. The six adult students who remained in the AH setting were enrolled in an English-Medium Education degree and received 20 h of instruction in English per week. Similarly to the children, the SA adults were exposed to naturalistic input both in and outside school, while the AH adults only received L2 input in class, as none of them reported using English outside of school.

3.2. Data Collection

The present study has a pre-/post-test design, and both tests were administered in the participants’ AH institution (in a school in the case of child participants, and in a university in the case of adults). The pre-test was administered a week prior to the participants’ departure to the target country, whereas the post-test was administered a week after their return from abroad. Although the young learners stayed abroad for two months while the adults enrolled in a three-month SA program, the intensity of L2 instruction was higher for the young learners (i.e., they had 25 h of class per week, while the adults received 8 to 12 h of instruction), hence allowing for comparable results between the two SA groups. All of the participants were interviewed individually in a quiet room by two researchers of the present study, and the interview was voice recorded.

The instrument used in this study was a semi-structured interview conducted in English. A semi-structured interview was included since this type of interview is an effective method for providing reliable, comparable data with different participants, even with different interviewers (Adams 2015). Additionally, semi-structured interviews (as opposed to other types of tasks such as monological picture-description tasks) allow participants to interact with their interlocutor and use free language while trying to keep content comparable. This semi-structured interview was previously used with similar participants (both children and adults) and was successful (Muñoz and Llanes 2014). The first questions of the interview were included to obtain basic information from the students and served as a warm-up. The rest of the questions were included because they were questions that all of the participants could answer without much cognitive effort, as questions requiring specific knowledge or cognitive effort would favor adults. Moreover, questions tackling different verbal tenses (present, past, and future) were included so that the complete final output was more authentic and included the use of PMs in cases where one verbal tense was not mastered.

The guiding questions asked to the participants are included in Appendix A. These involved some biographical questions that served as a warm-up, questions that required the use of several verbal tenses (i.e., what did you do yesterday, what will you do tomorrow?, what are your hobbies?, what would you like to do on your next holidays?), questions about experiences in their life (i.e., tell us the happiest moment in our life), and questions that required the use of different linguistic functions (i.e., describe your bedroom; ask me a few questions). The length of the interviews ranged from 2 min and 7 s to 6 min and 44 s, with an average length of 3.82 min for child participants, whereas the duration for the adult participants ranged from 4 min and 11 s to 11 min, with an average length of 6.93 min.

3.3. Data Coding and Analysis

The data were transcribed and coded using CLAN (MacWhinney 2000). More specifically, the pragmatic markers (PMs) elicited in the interviews were identified, considering previous studies that view PMs as placed on a functional continuum ranging from textual functions to interpersonal ones (Martín-Laguna 2020). For example, some PMs may be used with a textual function to continue the discourse (e.g., well), while others can be used with a more interpersonal function to relate to the interlocutor (e.g., you know). Some may play functions that are both interpersonal and textual; for instance, PMs used as fillers (e.g., um). Subsequently, we coded all PMs in English as L2PMs. Additionally, since the learners code-switched into Spanish and Catalan during the interviews, those PMs produced in the participants’ L1 (Spanish or Catalan) (L1PMs) were also included in the analysis.

Table 2. Pragmatic markers elicited in the present study.

L2PMs	L1PMs
Well, and, yeah, because, then, but, or, so I think, well, yeah, sorry, you know, yes	Bueno, vale, pues, sí, claro, buah, a veure, ai

presents the PMs elicited from the interviews in the two categories—L2PMs and L1PMs.

As shown in Table 2, the interviews elicited PMs with a range of different functions. Some of the functions were more textual, including signaling a topic shift of discourse (well, so), showing a continuation of discourse (and, yeah, so), marking disjunction (or, so), showing causal and consequential relationships (so), indicating results (so), or opening or closing the discourse (yeah, so). Other functions were more interpersonal, such as denoting a thinking process (I think, well), signaling receipt of information (yeah), showing repair (sorry), assessing the interlocutor’s knowledge (you know), and aligning oneself with the speaker (yes). Finally, the data included different L1PMs in Spanish (bueno, vale, pues, sí, buah and claro) and in Catalan (a veure, ai).

To illustrate how L2PMs and L1PMs were identified and coded, the excerpts below present sections of the interviews from the present study. Excerpt 1 is taken from an interview with one of the adult students, while Excerpt 2 is from an interview with a young learner.

Excerpt 1. Dialogue between interviewer (INV) and adult learner 1 (L01) in pre-test.

*INV: what did you do yesterday?

*L01: I didn’t study during Christmas. So [*L2PM] now, well [*L2PM], during the last week I had to study a lot.

As can be seen in Excerpt 1, L01 uses so as a PM to express consequence, while well is used as to denote a thinking process.

Excerpt 2. Dialogue between interviewer (INV) and learner 2 (L02) in pre-test.

*INV: and tomorrow? what will you do?

*L02: I I go bueno [*L1PM] hmm I will go to a to the church because and later I I’m going to go to my cousin’s house

In Excerpt 2, L02 uses the L1PM bueno (a Spanish PM) with the purpose of correcting herself.

To ensure the inter-rater reliability of the codification, one of the researchers of the study coded a random 10% of the data, and this was compared with the coding of the researcher who coded all of the transcriptions. The interrater reliability was 95.91%, and for the remaining 5%, any discrepancies were discussed, and agreement was reached.

Upon codification of the interview data, we calculated overall scores of PM use. To do so, the ratio of each type of PMs was calculated by multiplying the frequency of each PM per 100 words and dividing it by the number of words in the interview. This helped to control for differences in text length. Therefore, the formula to obtain the ratio of L2PM use was: number of L2PMs × 100/Tokens, while the formula to calculate L1PMs was: number of L1PMs × 100/Tokens. A higher number of L2PMs in the post-test would indicate that participants have a greater command of the L2, whereas for L1PMs, a higher score in the post-test would indicate the opposite. It must be pointed out that in order to compute the number of tokens (total words produced), the interventions of the investigator were excluded, and the data were pruned; that is, false starts, repetitions, unfinished sentences, and words in a language other than English were also excluded.

The data were analyzed quantitatively to answer the two RQs of the study. Preliminary analyses were conducted to confirm assumptions of normality of the data, linearity, homogeneity of variances, homogeneity of regression slopes, and reliable measurement of the covariate. Results from this preliminary testing, by means of a Kolmogorov-Smirnov test (p > 0.05), confirmed that the data were only normally distributed for the use of the PM and by children (p = 0.106). The rest of the data reported normality ratios with a significance value below 0.05 (p < 0.05), and therefore non-parametric tests were used. More particularly, we firstly conducted a series of Mann-Whitney U tests to address RQ1; that is, to explore differences between children and adults in the use of PMs in the pre-test (i.e., before the SA participants went abroad). Moving on to RQ2, Wilcoxon signed-rank tests were carried out to examine whether differences from pre- to post-test were significant across the different groups of participants (SA children, AH children, SA adults, and AH adults). Additionally, a further series of Mann-Whitney U tests were carried out to explore differences in PM gains (of L2PMs and L1PMs) between different groups of learners (SA children vs. AH children; SA adults vs. AH adults; SA children and SA adults; AH children and AH adults).

4. Results

4.1. Effects of Age on PM Use

The first RQ examined whether age (children vs. adults) affected the use of pragmatic markers (PMs). This analysis involved the pre-test interview data and compared the usage of L2 English PMs (L2PMs) and L1 Spanish and Catalan PMs (L1PMs) among the female children (n = 35) and adults (n = 16).

Table 3. Descriptive data on PM use in pre-test by children and adults.

	N	L2PMs M (SD)	L1PMs M (SD)
Children	35	2.10 (1.06)	0.66 (1.33)
Adults	16	5.13 (2.54)	0.19 (0.40)

provides a descriptive overview of this data.

As observed in Table 3, adults appeared to use a higher number of L2PMs (M = 5.13) compared to children (M = 2.10), whereas children seemed to utilize L1PMs more frequently (M = 0.66) than adults (M = 0.19). To assess the significance of these descriptive findings, a series of Mann-Whitney U tests were conducted. The results indicated significant differences between children and adults in the overall usage of L2PMs (U = 49.5; p < 0.001), favoring the adult group. However, no significant differences were found in the utilization of L1PMs between the two groups.

To further elucidate these findings, differences in the usage of each L2PM identified in the interviews were explored for both children and adults (see Table 2). Additional Mann-Whitney U tests revealed discrepancies between the two participant groups in their usage of the PMs I think (U = 115.5, p < 0.001), because (U = 43.5, p < 0.001), but (U = 55.5, p < 0.001), or (U = 146.5, p = 0.003), yes (U = 0, p < 0.001) and so (U = 94, p < 0.001). Specifically, adults employed all of these PMs more frequently than children, as depicted in

Table 4. Use of specific L2PMs in pre-test by children and adults.

PMs	Children (n = 35)	Adults (n = 16)
I think	0.60 (0.85)	2.50 (2.19)
Because	2.06 (1.37)	5.62 (2.39)
But	0.77 (1.11)	5.12 (4.30)
Or	0.51 (0.85)	1.94 (1.84)
Yes	0.06 (0.23)	10.00 (5.56)
So	0.31 (0.87)	4.19 (5.19)

, which presents the means and standard deviations of the PM frequency found in the corpus.

Following these results, Excerpts 3 and 4 exemplify the use of PMs by children (Excerpt 3) and by adults (Excerpt 4).

Excerpt 3 [child]

*INV: so you like learning English ‘name of the student’?

*L05: yeah.

*INV: why?

*L05: hmm because es (= is) hmm the most important language in the world so [*L2PM] hmm if I want to go to any place and I don’t know the language I can speak English.

Excerpt 4 [adult]

*INV: Can you describe your bedroom, please?

*L03: My bedroom? My bedroom is painted in blue. It’s plenty of things and [*L2PM] plenty of pictures. Isn’t really, is small, but [*L2PM], well [*L2PM].

Altogether, the results regarding RQ1 revealed that age appeared to influence the use of PMs, as the adults in the study utilized L2PMs more frequently than the children. Specifically, they employed the PMs I think, because, but, or, yes and so more frequently than the children. However, given that the children and adults had differing proficiency levels (children at B1 and adults at C1), these findings could also be attributed to the variable of proficiency.

4.2. Effects of Age and Context on the Development of PM Use

The second RQ investigated whether age (children vs. adults) and learning context (SA vs. AH) had any effects on the development of PM use. To address this question, the analysis examined differences among four groups: young learners enrolled in a two-month SA program in Ireland (n = 19), young learners who remained in their AH setting in Barcelona (n = 16), adults who participated in a three-month SA program in the UK or Ireland (n = 10), and adults who stayed in the AH context in Barcelona (n = 6).

Table 5. Descriptive data on PM use in pre-test, post-test and difference across age and context groups.

		L2PMs			L1PMs
	N	Pre	Post	Diff	Pre	Post	Diff
Children in SA	19	2.47 (1.15)	2.88 (1.33)	0.40 (1.41)	0.68 (1.41)	0.21 (0.71)	−0.47 (1.22)
Children AH	16	1.79 (0.69)	2.06 (1.47)	0.28 (1.26)	0.63 (1.25)	0.50 (1.03)	−0.13 (0.62)
Adults in SA	10	5.77 (2.40)	6.75 (3.26)	0.97 (2.27)	0.08 (0.29)	0.08 (0.29)	0.10 (0.32)
Adults AH	6	5.22 (3.09)	6.51 (1.13)	1.29 (1.61)	0.33 (0.52)	0.50 (0.84)	0.17 (0.75)

provides an overview of the descriptive data regarding their use of L2PMs and L1PMs in the pre-test (before the SA stays), post-test (after the SA stays), and the difference between pre- and post-test. Following the same format as Table 3, the figures indicate the means and standard deviations (in brackets).

As we can see in Table 5, adults in the AH context seemed to have experienced the largest gains in the frequency of L2PM use (M = 1.29), followed by the adults in the SA setting (M = 0.97), the children in the SA context (M = 0.40), and the children who remained AH showed the smallest gains (M = 0.28). Regarding the development of L1PMs, the children in SA contexts experienced the largest decrease in their use (M = −0.47), followed by the children AH (M = −0.13). However, the adults both in SA (M = 0.10) and AH (M = 0.17) settings slightly increased the use of L1PMs. To test the significance of these descriptive results, we first explored whether the difference between pre- and post-test was significant across the four groups, and in terms of L2PM and L1PM use. A series of Wilcoxon signed-rank tests revealed no significant differences in the overall use of L2PMs by any of the groups. Regarding the development of L1PMs, findings were significant for the children in the SA context (Z = −1.90; p = 0.048). That is, during the two months abroad, the children significantly reduced the number of L1PMs used. Nevertheless, the development of L1PMs was not significant for the rest of the groups.

Further Wilcoxon signed-rank tests were conducted to examine whether there were significant differences between pre- and post-test in the use of specific PMs (see Table 2). The results revealed that SA children experienced significant gains in the use of the L2PMs then (Z = 2.08; p = 0.038) [pretest = 1.95 (2.22); posttest = 3.95 (3.49); difference = 2.00 (3.81)] and yeah (Z = 1.88; p = 0.05) [pretest = 0.68 (1.60); posttest = 1.42 (1.46); difference = 0.74 (1.85)]. However, the children who stayed in the AH setting did not experience significant gains in the use of any of the L2PMs. Similarly, the adults who went abroad had significant gains in the use of the L2PM well (Z = 1.96; p = 0.049) [pretest = 5.60 (1.79); posttest = 3.40 (2.72); difference = 2.20 (2.70)], whereas no significant differences were reported for the adults who remained at home. These findings point to a key role of context in the development of L2PM use, as both children and adults in the SA setting increased their use of specific L2PMs, while the participants who remained at home did not experience changes in the use of any L2PM.

As can be seen in Excerpt 5, the child produces quite fluent and coherent discourse when answering the researcher’s question about the upcoming holidays. The participant uses well to begin her turn in the interview. Later on, she uses then to mark a series of events that will be taking place during the holidays, and finally, she uses “so” to provide an explanation and conclude her intervention.

Excerpt 5 [child]

*INV: and what would you like to do on your next holidays?

*L08: next holidays of this year or next year?

*INV: this year.

*L08: this year well [*L2PM] I don’t know because the thing that I’m going to do

• I am going to go to Formentera to two days no three and then [*L2PM]
• I’m going to Galicia because my the mother of my father lives there
• so [*L2PM] we’re going three weeks cause it’s the only time that I can see
• her so [*L2PM].

In the case of the adult learner, as can be seen in Excerpt 6, the PM well is used a couple of times in one turn. It is used together with filled pauses, both strategies used to gain time to think about the elaboration of her answer to the researcher’s question.

Excerpt 6 [adult]

*INV: what are your hobbies?

*L30: padel and I sing a lot in my house.

*INV: you sing a lot?

*L30: hmm well I love singing well hmm I don’t make classes but I love singing in my house.

Finally, we conducted a series of Mann-Whitney U tests to explore differences in L1PM and L2PM gains between different groups of learners (SA children vs. AH children; SA adults vs. AH adults; SA children and SA adults; AH children and AH adults). Although results regarding gains in L2PM use were not significant across the groups, some key significant findings were observed when the analysis focused on the development of specific L2PMs (see Table 2 above). More particularly, significant differences were reported for the development of the PM then between children in SA and children in AH (U = 96; p = 0.49). SA children experienced larger gains [M = 2; SD = 3.81] than children in AH settings [M = −0.06; SD = 1.77] in the use of this PM. This means that the SA context seems to be beneficial for the acquisition of then by children, given the possible exposure to this PM in a naturalistic setting. In contrast, perhaps the children in the AH setting were not exposed to this PM in their classroom input. Nevertheless, no significant differences were observed between the adults in the SA context and those AH.

As can be seen in Excerpt 7 below, the participant makes a correct use of the PM then to present a series of actions that were carried out in the past when asked what she did on the previous day.

Excerpt 7 [child]

*INV: have you been abroad to learn English?

*L10: yes twice last year to Ireland and this year again.

*INV: where?

*L10: to Ireland.

*INV: same place?

*L10: yeah.

*INV: hmm can you tell me what did you do yesterday?

*L10: yesterday I went to school hmm I woke up at like eight then I got a train down here and I did my Spanish exam I went lunch then came back and then I went to party to my friends’ party and I came back.

Significant differences were also observed between the children and the adults in the AH setting regarding the development of the L2PMs or (U = 23.5; p = 0.36) and yes (U = 23; p = 0.38). More particularly, the adults showed larger gains both in the PM or [M = 1.00; SD = 1.26] and yes [M = 3.83; SD = 4.99] than the children [or (M = −0.12; SD = 0.72); yes (M = 0.00; SD = 0.51)]. This means that there could be an effect of instruction in the foreign language (FL) setting, as perhaps the adults, unlike the children, were exposed to the PMs or and yes more frequently and in a more significant way in their classroom input. Regarding the SA context, however, no significant differences were found between the development of PM use by children and by adults. As for the L1PMs, significant differences were found between the children and the adults (U = 318.5; p = 0.011), since children in the SA setting significantly reduced the amount of L1s, unlike the adults in the same setting. Therefore, it may be claimed that, rather than context, it is age that mostly influenced the development of L1PM use across learning settings.

To illustrate this finding, Excerpt 8 shows how an adult AH learner uses the PM or twice in the same sentence to show contrast.

Excerpt 8 [adult]

*INV: Ok. You have a lot of work now, or what? Here at the university? Do you have lots of things to do?

*L11: Well [*IM], I have essays to hand in, to write or [*TM] just normal homework or [*TM] stuff like that.

In summary, findings related to RQ2 revealed a partial effect of age and context on the development of PMs. On the one hand, although no significant differences were found across age and context groups in the overall use of L2PMs, there were significant group differences regarding the development of the use of specific L2PMs. More particularly, the SA context seemed to be particularly beneficial for the increased use of the PMs then and yeah by children, whereas adults increased their use of well. Regarding the AH setting, the adults increased the frequency of use of the PMs or and yes. On the other hand, age played a role in the development of L1PM use in the SA context, as the children who went abroad significantly reduced the amount of L1PMs.

5. Discussion

The present study aimed to explore the development of PM use by children and adult L2 English learners in two learning contexts: SA and AH. A preliminary analysis pointed out that children and adults differed in their use of PMs, as adults used different types of L2PMs more frequently than children. However, we should also consider that proficiency could have played a role in this finding, as adults had a higher onset English proficiency level. Regarding the development of PM use across age and learning contexts, the findings revealed a partial effect of both factors on the observed differences between pre- and post-test. More particularly, the children who went abroad increased their use of the PMs then and yeah, whereas the adults who went abroad increased their use of well. As for the participants who stayed in the AH setting, only the adults experienced some pragmatic gains, as they increased the use of the PMs or and yes. Additionally, the children who went abroad, unlike the adults, reduced the amount of L1PMs used.

These findings are in line with previous longitudinal studies that have pointed out the beneficial role of SA for the development of PM use (De Cristofano and Badan 2021; Magliacane and Howard 2019; Magliacane 2020). As found in Achiba (2003) and Jones (2007), the participants of the present study increased the use of some PMs, as opposed to those who were AH. This increase may have been due to the amount of exposure and opportunities for interaction provided by the SA condition, which was not the case for those staying at home. The children in this study who went abroad on a two-month stay in Ireland increased their use of then and yeah, while the adults who went abroad for three months in the UK or Ireland increased their use of well. Moreover, the children who were abroad reduced the amount of L1PMs used. On the one hand, there are no previous studies about the development of PM use by children, and therefore this study starts a new line of research in this area of L2 pragmatics. On the other hand, the results regarding the adults corroborate previous studies by Magliacane and Howard (2019) and Magliacane (2020), who also focused on the acquisition of the PM well in a SA setting in Ireland (the adults of these study went on SA programs in either the UK or Ireland). As the authors illustrated, exposure to the L2 in a naturalistic context and regular contact with the target-language community enhances PM development in general, and the use of well in particular.

Regarding the findings related to the AH setting, the results from this study align with the majority of studies on PM use in CLIL/EMI settings, which have reported that young learners and adolescents in these settings underutilize PMs, mostly due to the lack of recurrent input containing PMs (Campillos-Llanos and Gozalo 2014; Hellermann and Vergun 2007; Lázaro Ibarrola 2016; Romero-Trillo 2002). However, in the case of adults, the results showed that those who remained in the AH setting increased their use of two PMs: or and yes. This finding suggests a possible effect of the quality of input received by adults in the EMI setting. As Lázaro Ibarrola (2016) explains, unlike traditional EFL learners, EMI learners are exposed to rich input in the L2 through both textbooks and classroom discourse, which may have enhanced their use of particular PMs. Indeed, this finding is consistent with previous studies that have observed that CLIL/EMI learners overuse specific PMs that are recurrent in textbooks and teachers’ speech (e.g., Hellermann and Vergun 2007).

6. Conclusions

In conclusion, the findings from the present study pointed to a partial effect of age (children vs. adults) and context (SA vs. AH) on the development of PM use. Children in the SA context increased their use of some L2PMs and reduced the amount of L1PMs used, whereas their AH counterparts did not show any pragmatic gains. Regarding the adults, those going abroad increased their use of one L2PM (well), while those who remained AH increased their use of two L2PMs (or and yes). With these findings, the study sheds some light on a topic that has not received much attention in the field of L2 pragmatics; that is, the development of PM use by children. Additionally, it contributes to the very small body of longitudinal studies that have explored the development of PM use in SA settings.

The reported findings have several pedagogical implications for L2 pragmatic learning in both SA and AH contexts in child and adult populations. First, as in previous studies discussed above, the current study shows that even a short SA experience may have positive effects on L2 pragmatic learning for both children and adults. In this line, school and university programs should promote such experiences since learners would be able to engage in real-life communicative situations with NSs and NNSs. Second, the fact that AH adult learners also improved in some PMs, but children did not, may suggest that children in AH contexts were less exposed to pragmatic exchanges than adult learners (maybe due to exposure/instruction in the classroom). In this line, EFL contexts should aim at including a wide variety of pragmatic moves in the class curriculum (across ages and proficiency levels), not only speech acts but also other pragmatic aspects such as PMs.

Like all research, the present study presents some limitations that may be addressed in future studies. Firstly, the small sample of participants has not allowed us to carry out a quantitative analysis based on parametric tests. Moreover, in an attempt to control the variable of gender, the sample only included females, and therefore the findings may not be generalized to the broader population that includes other genders. A second limitation involves the fact that the SA contexts of the study were limited to Ireland and the UK. Previous longitudinal studies on the development of English PM use in SA have also focused on Ireland. Therefore, more research is needed on a variety of English-speaking SA settings, including not only those where English is the predominant language (e.g., the United States, Australia, the United Kingdom) but also those where it is used as a second or additional language (e.g., Sweden, Malaysia, Nigeria). A further limitation included the fact that the two age groups had different onset proficiency levels, as expected when comparing children vs. adults. Future studies should aim to obtain two samples with the same proficiency level to make them more comparable. Finally, future studies should consider examining a wider range of tasks to assess oral production of PMs, as task type has been found to affect the use of PMs in children and adolescents.