Requests in Greek as a Foreign Language by Spanish/Catalan Bilinguals: The Role of Proficiency

Abstract

This study explores how Spanish/Catalan bilinguals acquire requests in Greek as a Foreign Language (FL), focusing on the role of proficiency in different communicative contexts. Fifty-four learners of Greek from different proficiency levels and fifty-three native Greek speakers participated in this study. Data was collected via role plays featuring varied social parameters (+/−Power, +/−Social Distance, +/−Imposition). Retrospective verbal reports were also employed to gain insights into learners’ use of requests, providing an overall view of their self-perceptions and pragmatic concerns across different proficiency levels. The findings revealed differences between native and non-native speakers in request types and the number of modifications, highlighting that increased proficiency does not necessarily result in target-like pragmatic performance. Additionally, social parameters clearly influenced learners’ requesting behavior, although their ability to interpret and appropriately respond to these variables developed inconsistently across different contexts and proficiency levels. Ultimately, the findings of this study may contribute to a better understanding of L2 pragmatic development in Greek as an FL and, in turn, inform pedagogical practices aimed at enhancing learners’ pragmatic competence.

1. Introduction

Several scholars have continuously emphasized the need for further research in Interlanguage Pragmatics (henceforth, ILP) from an acquisitional perspective (e.g., Alcón Soler & Martínez-Flor, 2008; Bardovi-Harlig, 1999; House & Kádár, 2023; Kasper & Rose, 2002), to analyze how L2 learners develop pragmatic competence (i.e., how to use language appropriately in different social situations) in different speech acts. Two key components of L2 pragmatic competence have been addressed in ILP: pragmalinguistic knowledge (i.e., linguistic resources for conveying meaning) and sociopragmatic knowledge (i.e., understanding social norms and context). However, studies often prioritize pragmalinguistic abilities over sociopragmatic ones, neglecting sociocultural factors that shape language use, interaction type, and interpersonal relationships (Alcón Soler & Martínez-Flor, 2008; Haugh et al., 2021; Taguchi, 2010). This oversight may hinder learners’ ability to communicate appropriately across diverse situations, underscoring the need to examine both pragmalinguistics and sociopragmatics in tandem (Usó Juan & Martínez Flor, 2008). In this sense, although with a different focus from acquisitional ILP studies, Kecskés’s (2013) theory of learners’ L1 cultural mindsets may help understand how they interpret social norms, perceive directness or politeness, and transfer pragmatic strategies from their native language when interacting in the L2. Similarly, recent ILP and language-pedagogy research shows that learners do not simply reproduce L1 norms or L2 input uncritically. Rather, they may exercise individual agency (Ishihara, 2019), selectively adopting, resisting, or negotiating pragmatic norms in ways that reflect personal preferences, identities, or communicative goals.

Although some studies suggest that certain aspects of pragmatics, such as requests, develop alongside language proficiency (Celaya & Barón, 2015; Rose, 2000), grammatical knowledge alone does not ensure pragmatic competence (Bardovi-Harlig, 1999; Félix-Brasdefer, 2003). Pragmatic development can plateau (Kasper & Rose, 2002), and even advanced learners “lack native-like sociopragmatic and pragmalinguistic knowledge, and lag behind in processing efficiency and fluency in pragmatic performance” (Taguchi, 2011, p. 909). Nonetheless, increased proficiency tends to support the expansion of pragmalinguistic repertoires (Zhang & Aubrey, 2024) and better alignment with sociopragmatic norms (Bardovi-Harlig, 2001; Kasper & Roever, 2005), thus highlighting the importance of both linguistic command and sociocultural awareness.

Among the various speech acts examined in ILP, requests—defined as “attempts by the speaker to get the hearer to do something” (Searle, 1979, p. 13)—have received considerable attention and remain one of the most extensively studied, due to its strong link with politeness and cross-cultural variation (e.g., Al Masaeed, 2022; Alcón Soler et al., 2005; Barón, 2015; Cañas, 2025; A. D. Cohen & Shively, 2007; Daskalovska et al., 2016; Gilabert & Barón, 2018; Khazdouzian et al., 2021; Lazarescu, 2021; Martínez-Flor, 2003; Savić et al., 2021; Taguchi, 2006). Research findings frequently highlight gaps in either learners’ pragmalinguistic or sociopragmatic competence. The latter aspect has been particularly examined in relation to the social parameters of Power (P), Social Distance (D), and Imposition (I), as proposed by Brown and Levinson (1987), which are believed to influence the degree of directness and mitigation strategies employed in requests. However, findings remain inconclusive. While some studies support the influence of these social parameters on learners’ request performance (e.g., Al-Gahtani & Roever, 2011; Alqurashi, 2022; Schauer, 2007; Seniarika et al., 2017), others suggest that learners often fail to adjust their requests according to contextual factors (e.g., Codina-Espurz, 2022; Economidou-Kogetsidis, 2008; Savić, 2015; Su & Ren, 2017; Göy et al., 2012). Notably, most ILP studies on requests focus on English as the target language, resulting in a significant gap in research on less commonly studied languages such as Greek. While the number of studies on requests involving Greek L1 speakers is increasing (e.g., Bella, 2025; Economidou-Kogetsidis, 2012, 2018, 2022; Koutsantoni, 2007), research on requests in Greek as an FL (Bella, 2012a, 2012b, 2014c) remains limited, which contributes to the continued underrepresentation of L2 pragmatics in instruction, particularly for Greek.

Greek culture has been described as being oriented toward positive politeness (e.g., Bella & Ogiermann, 2019; Pavlidou, 1994; Sifianou, 1992a, 1992b, 1999, 2023; Sifianou & Antonopoulou, 2005; Sifianou & Tzanne, 2010); it values politeness and formality, especially when making requests to elders, superiors, or strangers (see Florou, 2021), and this requires a high level of consideration for other people’s feelings. Given that reinforcing involvement and immediacy is important when making requests in Greek (Sifianou & Antonopoulou, 2005), the study of this speech act is essential for understanding how language reflects and maintains social norms and relationships in Greek society. Moreover, requests are often not perceived as face-threatening (Sifianou, 1992a). In fact, Bella (2025) identifies a “direct-yet-mitigated” request style in Greek in workplace interactions (e.g., Ξέρω ότι είστε πολύ απασχολημένος σήμερα, αλλά χρειάζομαι την αναφορά αν δεν σας δυσκολεύει/“I know you are very busy today, but I need the report, if it’s not too much trouble”). In this style, native speakers (NSs) employ syntactically direct expressions (e.g., want/need statements such as χρειάζομαι την αναφορά/“I need the report“) tempered by mitigating devices such as Disarmers or Considerators (e.g., Ξέρω ότι είστε πολύ απασχολημένος σήμερα/“I know you are very busy today,” or αν δεν σας δυσκολεύει/”if it’s not too much trouble”) that allow them to maintain politeness while asserting entitlement. While mitigating devices may sometimes be unnecessary (Sifianou, 1992a; Antonopoulou, 2001), this nuanced combination of directness and mitigation is characteristic of Greek request strategies (as in Bella, 2025) and may present challenges for L2 learners whose L1 pragmatics view directness as impolite.

Over the past few decades, a growing body of research in ILP has explored various speech acts in Greek, including apologies (Bella, 2014a), compliments (Sifianou, 2001), favor asking (Harissi, 2005), invitations (Bella, 2009), offers (Bella, 2016, 2019), refusals (Bella, 2009, 2011, 2014b), and thanking (Gkouma et al., 2020, 2023; Gkouma, 2024). Among these, requests have also received attention (Bella, 2012a, 2012b, 2014c; Economidou-Kogetsidis, 2002, 2005; Vassilaki & Selimis, 2020). However, research on Greek requests—particularly from a comparative and acquisitional perspective—remains limited, especially in FL contexts.

Economidou-Kogetsidis (2002, 2005) conducted several studies comparing Greek and English request strategies in professional settings, such as airline call centers. Findings showed that Greek speakers preferred more direct request forms, reflecting cultural norms prioritizing efficiency and spontaneity. Notably, this directness, while potentially face-threatening in English, was interpreted in Greek as a form of positive politeness. Moreover, directness sometimes served to increase rather than minimize Social Distance, challenging Brown and Levinson’s (1987) politeness theory.

Research into requests in Greek as an L2 is also limited. Bella (2012a) investigated how interaction with NSs and length of residence influence pragmatic development, focusing specifically on the use of modification devices in L2 learners’ requests. Her study compared two learner groups: one with a longer residence in Greece but limited interaction with NSs, and another with a shorter residence and more frequent contact with NSs. The non-native speakers (NNSs) were advanced learners of Greek enrolled in the University of Athens’ Teaching Center of Greek as a Second Language, all of whom had successfully passed the University’s C1 proficiency examination. Data was elicited through a Discourse Completion Task (DCT) in both formal and informal situations. Her findings suggest that learners who engage more with NSs develop better modification strategies, although both groups of NNSs still underperformed in the use of lexical/phrasal modifiers compared to NSs.

To our knowledge, the only existing studies about requests in Greek in the FL context are those conducted by Bella (2012b, 2014c), both of which investigate how proficiency level influences request performance. Using DCTs, Bella (2012b) found that, as learners’ proficiency increased, there was a shift from Direct to Conventionally Indirect Requests, accompanied by a broader use of both internal and external modifiers across two informal situations (−P, −D) and one formal situation (+P, +D). Notably, intermediate and advanced learners encountered greater difficulties in the informal contexts than in the formal one. Lower-level learners appeared to rely heavily on L1-based or universal pragmatic strategies, which suggests that pragmatic competence may develop prior to grammatical competence. However, none of the learners achieved native-like performance, largely due to limited grammatical and lexical resources.

Similar trends were observed in Bella’s (2014c) follow-up study, which examined intermediate and advanced learners’ request production in Greek as an FL across two communicative contexts: one involving interlocutors of equal status (−P, −D) and another involving interlocutors of unequal status (+P, +D). Although the learners showed some developmental progress, significant differences were found between both learner groups and NSs in both scenarios, indicating continued challenges in achieving native-like performance. In particular, learners tended to rely heavily on Direct Request strategies—at times even more so than NSs in comparable situations—and their use of internal modification (e.g., Downtoners, Hedgers, Conditional Forms) remained limited relative to NS norms. These gaps were attributed to factors such as cross-linguistic influence and sensitivity to contextual variation. Nonetheless, advanced learners outperformed their intermediate peers in the unequal-status situation, particularly in their use of head acts and syntactic modifiers.

In light of the aforementioned issues, this study, which builds on previous research by Cañas (2025), aims to contribute to the field by analyzing request production in Greek by Spanish/Catalan bilinguals across different proficiency levels, using open role plays. Additionally, this study aims to offer new insights into the developmental trajectory of L2 pragmatic competence in a less explored language. To this end, the study intends to answer the following research questions:

• What is the role of proficiency in the production of requests (head acts and modifications) in Greek as an FL by Spanish/Catalan bilinguals?
• How do sociopragmatic variables (Power, Social Distance, and Imposition) affect the production of requests in Greek as an FL by Spanish/Catalan bilinguals across different proficiency levels?

2. Materials and Methods

This study adopts a mixed-methods design, integrating both qualitative and quantitative approaches to provide a comprehensive analysis of the acquisition of the target item under investigation. In this study, we employed an explanatory sequential mixed-methods design (see Riazi & Candlin, 2014), in which a quantitative analysis of request realizations was conducted first and subsequently complemented by a qualitative examination of learners’ perceptions of the role plays. The qualitative phase was used to interpret and expand upon the quantitative patterns, providing a more comprehensive understanding of how advanced L2 learners produce requests in Greek.

2.1. LETEGR2 Project

This study forms part of the LETEGR2 project1 (Learning, Teaching, and Learning to Teach in Greek as a Second/Foreign Language: Evidence from Different Learning Contexts), coordinated by the National and Kapodistrian University of Athens in collaboration with various European institutions. Using a subset of the LETEGR2 corpus, specifically, role plays and retrospective verbal reports, the present study investigates how Spanish/Catalan bilinguals in Barcelona, Spain, perform requests in Greek as an FL across four proficiency levels, in various communicative situations involving different combinations of social parameters.

2.2. Participants

A total of 107 participants, recruited as part of the LETEGR2 project from two Greek language schools in Barcelona, took part in the main study (see Andria, 2024 for detailed descriptions of each school): 53 NSs aged 20–64 (M = 35.5, SD = 12.51) residing in Greece (used as a baseline group) and 54 Spanish/Catalan bilingual learners of Greek from the two language schools (NNSs). The NNS group, aged 22–78 (M = 50.5, SD = 16.08), showed diverse educational and linguistic backgrounds.

Learners’ proficiency levels were determined according to each language school’s internal categorization system, which is shared across both institutions. Students were placed into four proficiency levels as per the Common European Framework of Reference for languages (CEFR): A2, B1, B2, and C1. In these schools, learners typically begin at level A1 and progress sequentially through the curriculum. Incoming students who do not enter at the beginner level complete a placement test administered by the schools to determine their appropriate starting level.2 

Table 1. Number of non-native participants per level at both language schools.

Level	No of Participants	Language School 1		Language School 2
		Male	Female	Male	Female
A2	15	3	6	3	3
B1	16	4	4	5	3
B2	19	10	4	1	4
C1	4	0	0	0	4
Total	54	17	14	9	14

illustrates the distribution of participants across the different language CEFR-based proficiency levels.

Nearly half of the participants had prior exposure to Greek through travel or earlier coursework, and many had attended both schools at different times. Almost all participants held bachelor’s degrees, with some of them in humanities (linguistics, philology, and translation). They also spoke multiple foreign languages. Many had also visited Greece or studied Greek previously, indicating a high level of language learning motivation and engagement.

Prior to the data collection, all the participants provided written consent for their participation in the study. All dialogs were recorded with the participants’ knowledge. The project’s methodological design, as well as its research and ethics protocol, were approved by the review boards of the funding institutions (Project Code: 1656). In the institution where the project was carried out, a subsequent agreement was formalized, which includes methodological and ethical considerations (registration number: 037573/2018, 11/05/2018).

2.3. Instruments

In order to validate the research instruments employed in the present study, a pilot study was first conducted by the LETEGR2 research group involving 40 participants (10 NSs and 30 NNSs), which showed strong reliability (Cronbach’s α = 0.801).

2.3.1. Background Questionnaire

The background questionnaire, developed within the LETEGR2 project, was designed to collect essential participant information (biodata, language experience, and motivations, among others) and to identify the Spanish/Catalan bilinguals learning Greek as an FL for inclusion in the analysis. Since the data was collected in Barcelona, Spain, only participants meeting this criterion were included. The decision to focus on this specific L1 group was motivated by the fact that the present study forms part of a broader project that seeks to conduct cross-linguistic comparisons among Spanish, Catalan, and Greek. Consequently, controlling for participants’ L1 was essential to ensure a homogeneous and methodologically sound sample. Moreover, the inclusion of Spanish/Catalan bilinguals addresses an underrepresented population in research on L2 Greek pragmatics, thereby enriching the field with greater typological and sociolinguistic diversity. The participants could complete the questionnaire in Spanish, Catalan, or Greek, depending on their language preference, ensuring accessibility and accuracy in self-reporting.

2.3.2. Role Plays

The second instrument used in this study was three open role plays focusing on differing social parameters (+/−Power, +/−Social Distance, and +/−Imposition) (see Appendix A), originally developed within the framework of the LETEGR2 project, which included a total of ten role plays (seven target role plays and three distractors).

The studies conducted on requests in Greek so far have implemented the DCT as the main instrument for data collection, which presents certain limitations for measuring L2 pragmatic competence (see Beebe & Cummings, 2006; A. D. Cohen, 1996a; Demeter, 2007; Hartford & Bardovi-Harlig, 1992; Hinkel, 2005). However, as pointed out by several researchers (Alcón Soler & Martínez-Flor, 2008; Bardovi-Harlig, 2013; Beltrán-Palanques, 2020; Chang, 2006; Félix-Brasdefer, 2010, 2018; Houck & Gass, 2006; Kasper & Dahl, 1991), role plays present several advantages when investigating the development of pragmatic competence. Hence, the role play was the instrument chosen by the LETEGR2 project.

The role plays selected for the present study are the following:

• Role Play 1: Cleaning Scenario (−P, −D, +I). A person asks their roommate to clean the apartment after hosting a party.
• Role Play 2: Shoes Scenario (−P, +D, −I). A person asks a store employee to bring them a pair of shoes they wish to buy.
• Role Play 3: Deadline Extension Scenario (+P, +D, +I). A student asks their professor for an extension to submit an assignment.

2.3.3. Retrospective Verbal Reports

From a qualitative perspective, the third instrument used in this study was a complementary retrospective verbal report (see Appendix B), developed as part of the LETEGR2 project (Gkouma, 2024; Gkouma et al., 2020, 2023). These reports provided additional insights into participants’ prior experiences with the scenarios, their perceptions of the difficulty of each interaction, and their self-evaluations of performance. They also shed light on learners’ sociocultural understanding, cognitive processing, and familiarity with various communicative contexts (A. D. Cohen, 1996b; Félix-Brasdefer, 2010). It should be noted that the NNSs’ retrospective verbal reports did not focus specifically on the speech act of requests but had a more general scope. This was because the L2 data collection for the present study was part of broader data collection within the LETEGR2 research project, examining various aspects of Greek language acquisition. Participants could respond in Spanish, Catalan, or Greek.

2.4. Procedure

2.4.1. Data Collection

Data collection was carried out at the beginning of the 2019–2020 academic year and involved both NNSs and NSs. All participants gave written consent before the process began. The background questionnaire was first administered to elicit demographic and linguistic information.

Then, the role plays were collected individually in a quiet room with a trained researcher. Each participant received instructions, performed scenarios presented on cards one at a time in a random order, and was encouraged to act naturally. Scenarios included concise contextual information about social dynamics and were accompanied by visual aids to facilitate comprehension (Nguyen, 2019). Dialogs were audio-recorded with participants’ consent.

Lastly, retrospective verbal reports followed the role plays, capturing participants’ reflections. Each report lasted about three minutes on average, with most NNSs participating; four participants provided incomplete or unavailable responses. Participants could respond in Greek or their L1 to accommodate language proficiency.

The NS group followed a similar procedure, with most data collected in person in Athens, Greece. However, due to the onset of the COVID-19 pandemic, some sessions were conducted online via Zoom or Skype. Even though this shift only affected part of the NS sample, such differences do not seem to impact the results, as has recently been suggested by McManus et al. (2025). This change in procedure did not appear to negatively affect participants’ performance, as they maintained visual contact and accessed the scenario cards in the same way as their NNS peers. Call quality also remained consistently excellent.

2.4.2. Data Coding and Analysis

Following data collection, the data was coded based on Cañas’s (2025) coding scheme, which builds upon previous frameworks (e.g., Bella, 2012a, 2012b; Blum-Kulka & Olshtain, 1984), including Direct, Conventionally Indirect (CI), and Non-Conventionally Indirect (NCI) requests, as well as external and internal modifications. This scheme was selected because it allows for a detailed examination of both the request strategy (head act) and the modification devices learners use to mitigate or intensify the request. All analyses were based on the transcribed role-play interactions. Each request was coded for both head act strategies (degree of directness) and modification strategies (external and internal modifiers). It is important to note that a single request could include multiple modification strategies and that each identified strategy was coded separately.

Request strategies (head acts) represent the core linguistic realization of the request and vary in directness (e.g., Direct, CI, and NCI).

Table 2. Request categorization employed in the present research.

Degree of Directness	Strategy Types	Examples
Direct Request	Mood Derivable (Imperative)	Πλύνε τα ρούχα αμέσως./Wash the clothes immediately.
	Performative	Σας ζητάω ένα τσιγάρο./I am asking you for a cigarette.
	Obligation Statement	Πρέπει να πληρώνεις το λογαριασμό./You must pay the check.
	Need/Want Statement	Χρειάζομαι αύξηση μισθού./I need a pay raise.
Conventionally Indirect Request	Query Preparatory—Permission	Θα μπορούσα να δανειστώ το μολύβι σου;/Could I borrow your pencil?
	Query Preparatory—Ability	Μπορείτε να μου δώσετε άδεια για διακοπές;/Can you grant me vacation leave?
	Query Preparatory with Present Indicative (no modal)	Ετοιμάζεις εσύ τον καφέ και τον πίνουμε όταν είναι έτοιμος;/(Do) you prepare [present ind.] the coffee and we’ll drink it when it’s done?
	Suggestory Formula	Δεν βάζεις τα ρούχα στην ντουλάπα;/(Why) don’t you put the clothes in the closet?
Non-Conventionally Indirect Request	Hint	Τα έπιπλα είναι πολύ σκονισμένα./The furniture is very dusty.

Note. Based on Cañas (2025).

illustrates the coding scheme used for the head acts (main request) in the present study.

Regarding request modifications, external modifiers are supportive moves that occur outside the head act, such as Grounders, Preparators, or Apologies.

Table 3. External Modifications.

Types	Examples
Attention Getter	Συγγνώμη/Excuse me, Άκουσε/Listen, Να σου πω…/Let me tell you, Λοιπόν/So…, etc., and proper names.
Preparator	Θα ήθελα να σας ζητήσω μια μεγάλη χάρη./I’d like to ask you for a big favor.
Getting a Precommitment	Moυ κάνεις μία xάρη;/Can you do me a favor?
Grounder	Σιδέρωσε το πουκάμισο γιατί έχω πάρτι σήμερα το βράδυ./Iron the shirt because I have a party tonight.
Disarmer	Ξέρω ότι είσαι κουρασμένη από το ταξίδι αλλά πρέπει να ξεπακετάρουμε τις βαλίτσες./I know you’re tired from the trip, but we have to unpack the suitcases.
Imposition Minimizer	Θα ήθελα να ζητήσω άδεια, αλλά μόνο για δύο ημέρες./I’d like to ask for leave but just for two days.
Promise of Future Action/ Reward	Θα σας ενημερώσω σύντομα./I’ll let you know soon.
Apology	Συγγνώμη που το ζητάω, αλλά ξέχασα τις φωτοτυπίες στο σπίτι./I apologize for asking, but I forgot the photocopies at home.
Considerator	Aν έχεις βέβαια χρόνο, αλλιώς να πάω στο μαγαζί./If you have time of course. Otherwise, I’ll go to the store.
Discourse Orientation Move	Ξέρεις το τεστ που δίνω στις 15 του μήνα…/You know the exam I’m taking on the 15th…
Sweetener	Είσαι πάντα πολύ ευγενικός, μπορείς να με βοηθήσεις με αυτό το πρόβλημα;/ You’re always very kind, can you help me with this problem?

Note. Based on Cañas (2025).

below presents the coding scheme used in the present study for analyzing the types of external modifications employed by the participants in the role plays.

Internal modifiers are linguistic elements embedded within the head act, including syntactic mitigation (e.g., Conditional Forms, Past Tense, Passive Voice, etc.) or lexical/phrasal devices (e.g., Downtoners, Consultative Devices, etc.).

Table 4. Internal modifications.

Types		Examples
Syntactic Downgraders	Negation	Δεν θα μπορούσες να μου δώσεις λίγο χρήματα;/Couldn’t you give me some money?
	Subjunctive	Μήπως να στείλεις εσύ το γράμμα;/(Would you) maybe send [subjunctive] the letter?
	Conditional Structures	Θα ήθελα μία άδεια αν είναι δυνατόν./I would like a leave if it’s possible.
	Past Tense	Ήθελα να σας ζητήσω ένα δάνειο;/I wanted to ask you for a credit.
	Future Tense	Θα χρειαστώ ένα στυλό./I’ll need a pen.
	Aspect	Aναρωτιέμαι αν θα μπορούσα να έρθω μαζί σου στο σπίτι./I wonder if I can go back home with you.
	Passive Voice	Το σπίτι πρέπει να καθαριστεί./The house needs to be cleaned.
Lexical/Phrasal Downgraders	Understaters/Hedgers	λίγο/a little, κάπως/a bit
	Politeness Marker	παρακαλώ/please
	Subjectivizers	φοβάμαι/I’m afraid, νομίζω/I think, θεωρώ/I reckon, φαντάζομαι/I guess
	Downtoners	ίσως/perhaps, μήπως/maybe, απλώς οr απλά/just
	Cajolers	ξέρεις/you know, καταλαβαίνεις/you understand
	Solidarity Markers	ρε, μωρέ/dude, diminutives, affectionate terms, person’s name + possessive pronoun (Ελένη μου/My Helen), αγόρι μου/my boy, κορίτσι μου/my girl
	Consultative Devices	νομίζεις ότι θα μπορούσα…/do you think I could…, θα ήταν δυνατόν/would it be possible…, είναι εντάξει αν/is it ok if…
	Appealers	εντάξει;/ok? έτσι δεν είναι; isn’t it? Ναι;/right?
Upgraders	Intensifiers	απολύτως/absolutely, εντελώς/completely, εξαιρετικά/extremely, αρκετά/rather, πραγματικά/really, τόσο/so, πάρα πολύ/totally, καθόλου/at all, αυτή τη στιγμή/right now, τώρα/now, αμέσως/immediately.—Πραγματικά πρέπει να κάνετε την εργασία σας./You really must do the homework.

Note. Based on Cañas (2025). The internal modification of “Aspect” (Blum-Kulka et al., 1989), indicates uncertainty when formulating a request (e.g., “I wonder”). This Syntactic Downgrader will be maintained in the present study for Greek requests. Thus, it should not be mistaken for the grammatical aspect of Greek verbs (i.e., complete vs. incomplete action). Regarding the labeling of “please,” it sometimes functions solely as a politeness marker, while in other contexts—particularly when followed by the particle σας (formal “you”) as in Example 6d—it conveys both politeness and solidarity, signaling respect and alignment with social norms.

below shows the coding scheme for internal modification employed by the participants in the present study.

To ensure reliability in the coding process, 20% of the data was independently coded by two of the three authors of this study. Cohen’s Kappa was used to assess interrater agreement. The Cleaning Scenario showed substantial agreement (κ = 0.718, 87%), while the Shoes Scenario yielded perfect agreement (100%, κ = NaN due to no variability). The Deadline Extension Scenario demonstrated almost perfect agreement (κ = 0.871, 91.3%). Overall, the observed agreement across the coded data was 92.77%. Any discrepancies were resolved through discussion and consensus.

Descriptive statistics were used to summarize frequencies, percentages, means, and standard deviations of request strategies and modifications across groups. Additionally, inferential statistics were conducted through Nominal Logistic Regression tests to assess the influence of proficiency on the likelihood of using specific head act types, and one-way ANOVA and Post Hoc Tests were performed to examine whether the number of modifications differed significantly between groups. Effect sizes for the one-way ANOVA were calculated using eta-squared (η²), with benchmarks of 0.01, 0.06, and 0.14 representing small, medium, and large effects, respectively (J. Cohen, 1988).

Despite non-normal distribution results from the Shapiro–Wilk test (p < 0.05), parametric tests were retained due to their robustness and support from prior studies (e.g., Blanca et al., 2017, 2023). Visual checks (via histograms and Q-Q plots) also indicated no severe violations of normality.

All statistical analyses were conducted using SPSS version 29.

Regarding the retrospective verbal reports, they were transcribed verbatim and analyzed qualitatively. More specifically, a deductive thematic analysis was employed (Braun & Clarke, 2006). The codes were predetermined by the questions that were the foci of the verbal report (learners’ familiarity and prior experience with the situations and the level of difficulty they perceived in each interaction). Participants’ responses to the same questions were then compared across groups to identify similarities and differences. These qualitative findings were then triangulated with the quantitative results, providing complementary insights into performance differences across scenarios.

3. Results

This section will present the main findings concerning the participants’ acquisition of requests in Greek while focusing on proficiency and the effect of varied social parameters.

3.1. Cleaning Scenario (−P, −D, +I)

3.1.1. Head Acts

Figure 1 presents the results of the three main types of head acts used by NSs and each proficiency level of NNSs in this particular scenario.

Table 5. Overall distribution of requests used by NSs and NNSs per level in the Cleaning Scenario.

	NS		A2		B1		B2		C1
Request Type (Head Act)	f	%	f	%	f	%	f	%	f	%
Direct
Mood Derivable (Imperative)	3	6	0	0	0	0	1	7	0	0
Performative	0	0	1	6	1	6	0	0	0	0
Obligation Statement	8	16	2	13	4	26	8	57	1	25
Need/Want Statement	1	2	1	6	0	0	0	0	0	0
Total Direct	12	24	4	26	5	33	9	64	1	25
Conventionally Indirect
Query Preparatory—Permission	0	0	0	0	0	0	1	7	0	0
Query Preparatory—Ability	10	20	5	33	4	26	1	7	3	75
Query Preparatory with Present Indicative (no modal)	1	2	1	6	2	13	0	0	0	0
Suggestory Formula	7	14	1	6	1	6	1	7	0	0
Total Conventionally Indirect	18	36	7	46	7	46	3	21	3	75
Non-Conventionally Indirect
Hint	20	40	3	20	3	20	2	14	0	0
Total Non-Conventionally Indirect	20	40	3	20	3	20	2	14	0	0
Unclear/Incomplete Requests	0	0	1	6	1	6	0	0	0	0
Total Head Acts	50	100	15	100	15	100	14	100	4	100

includes the overall distribution of requests employed by NSs and NNSs in this scenario. From the total cohort, data for this role play were unavailable for three NSs and six NNSs, specifically one at the B1 level and five at the B2 level.

As shown in Figure 1 and further detailed in Table 5, there were several differences in the use of requests across the groups. NSs mostly preferred to use NCI Requests (n = 20) followed by CI Requests (the Query Preparatory of Ability) (n = 10), and Direct Requests (Obligation Statement) (n = 8). NNSs mostly favored the Query Preparatory of Ability (A2: n = 5; B1: n = 4; C1: n = 3); however, most participants at the B2 level opted for Direct Requests using the Obligation Statement (n = 8), which was barely used by the NSs. Hints (n = 3) were only used by some NNS participants, primarily at the lower levels (A2 and B1). The following examples highlight this variation in the types of requests used across groups in the Cleaning Scenario:

(1)	a.	A2. Το σπίτι είναι βρόμικο. [Non-Con./Hint]
		Eng. The house is dirty.
	b.	B1. Aλλά δεν είναι καλά που όλα είναι βρόμικο. [Non-Con./Hint]
		Eng. It’s not good that everything is dirty.
	c.	B2. Πρέπει να το καθαρίσεις. [Direct/Obligation Statement]
		Eng. You must clean it.
	d.	C1. Μπορείς εσύ να μαζέψεις λίγο τα πράγματα και αν χρειάζεται αύριο μπορούμε να το τελειώσουμε μαζί; [Query Preparatory of Ability]
		Eng. Could you tidy up a bit, and if needed, we can finish it together tomorrow?
	e.	NS. Το σπίτι είναι λίγο βρόμικο από χτες. [Non-Con./Hint]
		Eng. The house is a bit dirty since yesterday.

Table 6. Means and standard deviations of the use of requests by NSs and NNSs in the Cleaning Scenario.

Groups		Direct	Conventionally Indirect	Non-Conventionally Indirect
NS	Mean	3	4.5	20
	Standard Deviation	3.8	4.15	0
A2	Mean	1	1.75	3
	Standard Deviation	0.71	1.92	0
B1	Mean	1.25	1.75	3
	Standard Deviation	1.64	1.48	0
B2	Mean	2.25	0.75	2
	Standard Deviation	3.35	0.43	0
C1	Mean	0.25	0.75	0
	Standard Deviation	0.43	1.30	0

below shows the means and standard deviations of the use of the types of requests in the Cleaning Scenario.

A nominal logistic regression3 was conducted to examine the effect of proficiency level on the probability of using Direct, CI, or NCI Requests in the Cleaning Scenario. Overall group differences were not statistically significant (Wald = 10.57, p = 0.103). However, significant pairwise differences were obtained between the B2 group and NSs. Specifically, the odds of producing a CI Request (vs. a Direct Request) were significantly lower for B2 participants compared to NSs (Exp(b) = 0.22, p < 0.05). Similarly, the odds of producing an NCI Request (vs. a Direct Request) were also significantly lower for the B2 group (Exp(b) = 0.13, p < 0.05).

3.1.2. Modifications

Regarding modifications, the frequencies of their use by the different groups of participants are displayed in Figure 2.

As observed in the above figure, all groups exhibit a higher preference for external modification. However, NSs used more external (n = 99) and internal (n = 72) modifiers overall in this particular scenario. In contrast, all groups of NNSs employed fewer modification devices, lagging far behind their NS peers. Regarding the external modifiers, B2 participants employed them considerably (n = 35), followed by participants at the A2 and B1 levels, who used the same number of devices (both n = 28). C1 participants mostly used external modifications (n = 12). Conversely, the highest number of internal modifiers (n = 14) was employed by the B2 group.

Table 7. Means and standard deviations of the use of modifications by NSs and NNSs in the Cleaning Scenario.

Groups		External	Internal	Total
NS	Min	0	0	0
	Max	5	3	7
	Mean	1.87	1.38	3.25
	Standard Deviation	1.35	1.06	1.82
A2	Min	0	0	0
	Max	3	1	3
	Mean	1.87	0.27	2.13
	Standard Deviation	0.91	0.45	0.91
B1	Min	0	0	1
	Max	3	2	4
	Mean	1.75	0.38	2.13
	Standard Deviation	1.00	0.71	1.08
B2	Min	0	0	0
	Max	11	2	12
	Mean	1.84	0.74	2.58
	Standard Deviation	2.60	0.80	3.00
C1	Min	2	0	2
	Max	5	5	8
	Mean	3.00	1.50	4.50
	Standard Deviation	1.41	2.38	3.00

includes the means and standard deviations of external and internal modifications used by each group of participants.

A one-way ANOVA revealed no statistically significant differences between groups in the number of request modifications in the Cleaning Scenario (F(2, 381) = 4, p = 0.056). The effect size was small, with an eta squared (η²) of 0.021. The following examples show some modifications used by one learner at the C1 level and a NS:

(2)	a.	C1. Πωπωπωπω, πολύ βρόμικο είναι. [Grounder] Κατερίνα [Attention Getter] τι έκανες εδώ; [Discourse Orientation Move] Ναι, αλλά όλα είναι βρόμικα. [Grounder] […] Κοίτα εγώ είμαι πτώμα, είμαι πολύ κουρασμένη, δεν μπορώ να σε βοηθήσω τώρα. [Grounder] μπορείς εσύ να μαζέψεις λίγο [Understater] τα πράγματα και αν χρειάζεται αύριο [Considerator] μπορούμε να το τελειώσουμε μαζί;
		Eng. Wow, it’s so dirty. [Grounder] Katerina, [Attention Getter] what did you do here? [Discourse Orientation Move] Yeah, but everything’s a mess. [Grounder] Look, I’m exhausted, I’m really tired, I can’t help you right now. [Grounder] Can you… clean up a bit [Understater], and if you need it, [Considerator] we can finish it together tomorrow?
	b.	NS. Να σου πω, [Attention Getter] νομίζω ότι [Subjectivizer] μετά τα χθεσινά πρέπει λίγο [Understater] να συμμαζέψουμε και να επικρατήσει η καθαριότητα στον χώρο. [Grounder]
		Eng. Listen, [Attention Getter] I think [Subjectivizer] after everything that happened yesterday, we should tidy up a bit [Understater] and bring some cleanliness back to the place. [Grounder]

Table 8. Overall distribution of external modifications used by NSs and NNSs in the Cleaning Scenario.

External Modifications		Cleaning
		NS	A2	B1	B2	C1
Attention Getters	f	24	6	3	3	1
	%	24	21	10	8	8
Preparators	f	5	1	3	3	0
	%	5	3	10	8	0
Getting a Precommitment	f	0	0	0	0	0
	%	0	0	0	0	0
Grounder	f	41	11	10	22	6
	%	41	39	35	62	50
Disarmer	f	1	0	2	0	1
	%	1	0	7	0	8
Imposition Minimizer	f	0	0	0	0	0
	%	0	0	0	0	0
Promise of Reward/Future Action	f	0	0	0	0	0
	%	0	0	0	0	0
Apology	f	0	0	0	0	0
	%	0	0	0	0	0
Considerator	f	2	0	0	1	1
	%	2	0	0	3	8
Discourse Orientation Move	f	26	10	10	6	3
	%	26	35	35	17	25
Sweeteners	f	0	0	0	0	0
	%	0	0	0	0	0
Total	f	99	28	28	35	12
	%	57	87	82	71	66

displays the distribution of the types of external modifiers used by NSs and all NNS groups in this role play.

As can be observed, both NSs and NNSs mostly opted for the same types of external modifications to accompany their main request in this scenario: Attention Getters, Grounders, and Discourse Orientation Moves. Despite this similarity, the use of these three modifiers by NNSs seemed to decrease with proficiency. As far as the internal modifications are concerned,

Table 9. Overall distribution of internal modifications used by NSs and NNSs in the Cleaning Scenario.

Internal Modifications			Cleaning
			NS	A2	B1	B2	C1
Syntactic	Negation	f	1	0	1	0	0
		%	1	0	16	0	0
	Subjunctive	f	3	0	0	0	0
		%	4	0	0	0	0
	Conditional Structure	f	6	0	0	1	0
		%	8	0	0	7	0
	Past Tense	f	0	0	0	0	0
		%	0	0	0	0	0
	Future Tense	f	10	0	1	0	0
		%	13	0	16	0	0
	Aspect	f	0	0	0	0	0
		%	0	0	0	0	0
	Passive Voice	f	2	0	0	0	0
		%	2	0	0	0	0
Lexical/Phrasal	Understaters/Hedgers	f	14	0	0	0	2
		%	19	0	0	0	33
	Politeness Marker	f	1	2	1	3	2
		%	1	50	16	21	33
	Subjectivizers	f	3	0	0	3	0
		%	4	0	0	21	0
	Downtoners	f	10	0	0	2	0
		%	13	0	0	14	0
	Cajolers	f	0	0	0	1	0
		%	0	0	0	7	0
	Solidarity Markers	f	10	0	0	1	2
		%	13	0	0	7	33
	Consultative Devices	f	6	1	1	1	0
		%	8	25	16	7	0
	Appealers	f	3	0	0	0	0
		%	4	0	0	0	0
	Intensifiers (Upgraders)	f	3	1	2	2	0
		%	4	25	33	14	0
Total		f	72	4	6	14	6
		%	42	12	17	28	33

presents the distribution of modifiers used per group in this scenario.

As shown, the NS group mostly favored the Future Tense (n = 10) and Conditional Structure (n = 6) while also employing the Subjunctive (n = 3), Passive Voice (n = 2), and Negation (n = 1) to some extent. In contrast, NNSs scarcely used any of these devices to modify their requests internally. Additionally, NSs displayed a higher use of lexical/phrasal modifiers than did the NNSs. The modifiers mostly preferred by NSs in this scenario were as follows: Understaters/Hedgers (n = 14), Downtoners (n = 10), and Solidarity Markers (n = 10). In contrast, these modifiers were barely used by NNSs, who opted instead for a limited use of Politeness Markers, Consultative Devices, and Intensifiers.

3.2. Shoes Scenario (−P, +D, −I)

3.2.1. Head Acts

Figure 3 shows the distribution of the three main types of head act used by NSs and across learners’ proficiency levels in this scenario.

Table 10. Overall distribution of requests used by NSs and NNSs per level in the Shoes Scenario.

	NS		A2		B1		B2		C1
Request Type (Head Act)	f	%	f	%	f	%	f	%	f	%
Direct
Mood Derivable (Imperative)	0	0	0	0	0	0	0	0	0	0
Performative	0	0	0	0	0	0	0	0	0	0
Obligation Statement	0	0	0	0	0	0	0	0	0	0
Need/Want Statement	17	32	11	73	7	43	12	63	2	50
Total Direct	17	32	11	73	7	43	12	63	2	50
Conventionally Indirect
Query Preparatory—Permission	9	16	0	0	3	18	2	10	0	0
Query Preparatory—Ability	10	18	0	0	1	6	0	0	0	0
Query Preparatory with Present Indicative (no modal)	14	26	3	20	1	6	4	21	1	25
Suggestory Formula	0	0	0	0	0	0	0	0	0	0
Total Conventionally Indirect	33	62	3	20	5	31	6	31	1	25
Non-Conventionally Indirect
Hint	3	5	1	6	3	18	1	5	1	25
Total Non-Conventionally Indirect	3	5	1	6	3	18	1	5	1	25
Unclear/Incomplete Requests	0	0	0	0	1	6	0	0	0	0
Total Head Acts	53	100	15	100	16	100	19	100	4	100

below presents the overall distribution of requests employed by all participant groups.

While NSs strongly preferred CI Requests (n = 33), NNSs opted more for Direct Requests across all proficiency levels. Notably, Need/Want Statements (A2: n = 11; B1: n = 7; B2: n = 12: C1: n = 2) were the most favored among NNSs, and were also employed by NSs but to a lesser degree (n = 17). The use of CI Requests by the NNSs was scarce, with only some participants at the B2 level using them (n = 4). The following examples illustrate the use of request types among the groups:

(3)	a.	A2. Το σπίτι είναι βρόμικο. [Direct/Need-Want Statement]
		Eng. I want these shoes.
	b.	B1. Εγώ θέλω ένα *φεγάρι παπούτσια. [Direct/Need-Want Statement]
		Eng. I want a pair of shoes.
	c.	B2. Ήθελα να αγοράσω ένα ζευγάρι παπούτσια που είδα την περασμένη εβδομάδα. [Direct/Need-Want Statement]
		Eng. I wanted to buy a pair of shoes that I saw last week.
	d.	C1. Θα ήθελα να αγοράσω αυτό το ζευγάρι παπούτσια που έχετε εδώ. [Direct/Need-Want Statement]
		Eng. I would like to buy this pair of shoes that you have here.
	e.	NS. Μήπως έχετε αυτό το ζευγάρι παπούτσια σε 38; [Query Preparatory with Present Indicative]
		Eng. Do you have this pair of shoes in size 38?

The means and standard deviations for the use of requests by each group are shown in

Table 11. Means and standard deviations of the use of requests by NSs and NNSs in the Shoes Scenario.

Groups		Direct	Conventionally Indirect	Non-Conventionally Indirect
NS	Mean	4.25	8.25	3
	Standard Deviation	8.5	5.9	0
A2	Mean	2.75	0.75	1
	Standard Deviation	5.5	1.5	0
B1	Mean	1.75	1.25	3
	Standard Deviation	3.5	1.25	0
B2	Mean	3	1.5	1
	Standard Deviation	6	1.9	0
C1	Mean	0.5	0.25	1
	Standard Deviation	1	0.5	0

.

A nominal logistic regression was conducted to examine the effect of proficiency level on the probability of using Direct, CI, or NCI Requests in the Shoes Scenario. Overall group differences were statistically significant (Wald = 15.28, p = 0.018). Pairwise comparisons revealed significant differences between the A2 group and NSs. Specifically, the odds of producing a CI Request (vs. a Direct Request) were significantly lower for A2 participants compared to NSs (Exp(b) = 0.14, p < 0.05). Similarly, the odds of producing a CI Request (vs. a Direct Request) were also significantly different between the B2 group and NSs, Exp(b) = 0.26, p < 0.05.

3.2.2. Modifications

Figure 4 below displays the frequencies of the use of external and internal modifications by the different groups in this particular scenario.

As seen in Figure 4, there was a greater preference for external modifications over internal ones by all groups. NSs used more external (n = 62) than internal modifiers (n = 39), similarly to both B1 and B2 NNSs, who also used more external modifiers (both n = 20). The main difference between B1 and B2 participants was observed in their use of internal modifications (n = 7 and n = 13, respectively). Moderate usage of modifications was shown among participants at the A2 level (external: n = 9; internal: n = 2). Finally, participants at the C1 level made considerable use of modifications.

Table 12. Means and standard deviations of the use of modifications by NSs and NNSs in the Shoes Scenario.

Groups		External	Internal	Total
NS	Min	0	0	0
	Max	4	2	5
	Mean	1.17	0.74	1.91
	Standard Deviation	1.03	0.71	1.13
A2	Min	0	0	0
	Max	2	1	2
	Mean	0.60	0.13	0.73
	Standard Deviation	0.63	0.35	0.79
B1	Min	0	0	0
	Max	4	2	4
	Mean	1.25	0.44	1.63
	Standard Deviation	1.00	0.62	0.95
B2	Min	0	0	0
	Max	3	2	3
	Mean	1.05	0.68	1.74
	Standard Deviation	0.91	0.67	0.99
C1	Min	1	1	2
	Max	4	3	6
	Mean	2.25	1.75	4.00
	Standard Deviation	1.25	0.95	1.82

below displays the relevant means and standard deviations for both types of modifications used by all groups.

A one-way ANOVA revealed significant differences between groups in the number of modifiers in the Shoes Scenario (F(8, 135) = 4, p < 0.001). The effect size was large, with an eta squared (η²) of 0.242. Post hoc tests using Tukey’s HSD identified statistically significant differences between several proficiency levels: A2 and C1, (p < 0.001), A2 and NSs (p = 0.003), B1 and C1 (p = 0.001), B2 and C1 (p = 0.002), and C1 and NSs (p = 0.002). No statistically significant differences were revealed when comparing the B1 and B2 levels to each other or when they were compared to NSs. The examples below show the use of modifications by participants in different groups in this particular scenario:

(4)	a.	A2. Μου αρέσουν αυτά τα παπούτσια πάρα πολύ. [Discourse Orientation Move] θέλω να αγοράσω.
		Eng. I like these shoes a lot [Discourse Orientation Move]. I want to buy (them)
	b.	B1. Θα ήθελα [Conditional] να δω δύο *ζευγάρι παπούτσια, για *ένα γιορτή [Grounder].
		Eng. I would like to see two pairs of shoes, for a party [Grounder].
	c.	B2. Ήθελα [Past Tense] να αγοράσω ένα ζευγάρι παπούτσια που είδα την περασμένη εβδομάδα…
		Eng. I wanted [Past Tense] to buy a pair of shoes that I saw last week…
	d.	C1. Συγγνώμη [Attention Getter], έχω δει ένα ζευγάρι παπούτσια εκεί στο παράθυρο [Grounder] και δεν ξέρω [Subjectivizer], αν έχετε το νούμερό μου. Είναι το τριάντα έξι [Grounder].
		Eng. Excuse me [Attention Getter], I’ve seen this pair of shoes here in the store window and I don’t know [Subjectivizer] if you have my size. It’s thirty-six [Grounder].
	e.	NS. Συγγνώμη [Attention Getter], να σας ρωτήσω κάτι; [Preparator] υπάρχουν μήπως [Downtoner] σε 43; Γιατί τα βλέπω στη βιτρίνα μόνο μεγάλα… 45… [Grounder]
		Eng. Excuse me [Attention Getter], can I ask you something? [Preparator] do you happen to [Downtoner] have them in a size 43? Because I see them only big in the window case … 45… [Grounder]

Table 13. Overall distribution of external modifications used by NSs and NNSs in the Shoes Scenario.

External Modifications		Shoes
		NS	A2	B1	B2	C1
Attention Getters	f	17	2	4	2	2
	%	27	22	20	10	22
Preparators	f	20	0	0	0	0
	%	32	0	0	0	0
Getting a Precommitment	f	3	0	1	0	0
	%	5	0	5	0	0
Grounder	f	11	1	3	11	7
	%	17	11	15	55	77
Disarmer	f	0	0	0	0	0
	%	0	0	0	0	0
Imposition Minimizer	f	0	0	0	0	0
	%	0	0	0	0	0
Promise of Reward/Future Action	f	0	0	0	0	0
	%	0	0	0	0	0
Apology	f	0	0	0	0	0
	%	0	0	0	0	0
Considerator	f	1	0	0	0	0
	%	1	0	0	0	0
Discourse Orientation Move	f	10	6	12	7	0
	%	16	66	60	35	0
Sweeteners	f	0	0	0	0	0
	%	0	0	0	0	0
Total	f	62	9	20	20	9
	%	61	81	74	60	56

presents the distribution of external modifier types used by NSs and NNSs in this role play.

As shown, NSs mostly prioritized Preparators (n = 20), Attention Getters (n = 17), Grounders (n = 11), and Discourse Orientation Moves (n = 10) in this role play. In contrast, NNSs relied more heavily on Grounders (A2: n = 1; B1: n = 3; B2: n = 11; C1: n = 7) and Discourse Orientation Moves (A2: n = 6; B1: n = 12: B2, n = 7). As far as the internal modification devices are concerned,

Table 14. Overall distribution of internal modifications used by all groups in the Shoes Scenario.

Internal Modifications			Shoes
			NS	A2	B1	B2	C1
Syntactic	Negation	f	0	0	0	0	0
		%	0	0	0	0	0
	Subjunctive	f	0	0	0	0	0
		%	0	0	0	0	0
	Conditional Structure	f	19	2	3	6	2
		%	48	100	42	46	28
	Past Tense	f	2	0	1	2	1
		%	5	0	14	15	14
	Future Tense	f	0	0	0	0	0
		%	0	0	0	0	0
	Aspect	f	1	0	1	0	0
		%	2	0	14	0	0
	Passive Voice	f	1	0	0	0	0
		%	2	0	0	0	0
Lexical/Phrasal	Understaters/Hedgers	f	1	0	1	0	0
		%	2	0	14	0	0
	Politeness Marker	f	6	0	0	3	0
		%	15	0	0	23	0
	Subjectivizers	f	0	0	0	0	1
		%	0	0	0	0	14
	Downtoners	f	7	0	1	1	1
		%	18	0	14	7	14
	Cajolers	f	0	0	0	0	0
		%	0	0	0	0	0
	Solidarity Markers	f	0	0	0	1	0
		%	0	0	0	7	0
	Consultative Devices	f	3	0	0	0	2
		%	7	0	0	0	28
	Appealers	f	0	0	0	0	0
		%	0	0	0	0	0
	Intensifiers (Upgraders)	f	0	0	0	0	0
		%	0	0	0	0	0
Total		f	39	2	7	13	7
		%	38	18	25	39	43

shows the frequencies of the different types of syntactic and lexical/phrasal modifiers used across groups.

NSs showed a clear preference for the Conditional Structure (n = 19). Similarly, this modifier was also employed by some NNSs across proficiency levels, though to a lesser degree. Notably, its usage increased with proficiency, from A2 (n = 2) to B1 (n = 3) and B2 (n = 6), before declining again at the C1 level (n = 2). The B2 group exhibited the highest frequency of syntactic modifiers in their requests. Concerning the lexical/phrasal modifiers, Downtoners (n = 7) and Politeness Markers (n = 6) were the most frequently used lexical/phrasal modifiers by NSs. The NNSs at lower proficiency levels made less use of these types of modifiers in their requests. Once again, the B2 group used the highest number of modifiers overall in this scenario among the NNS groups. The C1 group also displayed some use of lexical/phrasal modifiers.

3.3. Deadline Extension Scenario (+P, +D, +I)

3.3.1. Head Acts

Figure 5 illustrates how the three main head act types are distributed across NSs and the learners’ proficiency levels in this scenario.

The overall distribution of request strategies employed by NSs and NNSs in this scenario are presented in

Table 15. Overall distribution of requests used by NSs and NNSs per level in the Deadline Extension Scenario.

	NS		A2		B1		B2		C1
Request Type (Head Act)	f	%	f	%	f	%	f	%	f	%
Direct
Mood Derivable (Imperative)	1	2	0	0	0	0	0	0	0	0
Performative	2	4	0	0	0	0	0	0	0	0
Obligation Statement	0	0	0	0	0	0	0	0	0	0
Need/Want Statement	17	32	3	20	5	31	6	33	1	25
Total Direct	20	37	3	20	5	31	6	33	1	25
Conventionally Indirect
Query Preparatory—Permission	17	32	8	53	6	37	8	44	1	25
Query Preparatory—Ability	13	24	2	13	2	12	3	16	2	50
Query Preparatory with Present Indicative (no modal)	0	0	1	6	1	6	0	0	0	0
Suggestory Formula	2	4	0	0	0	0	0	0	0	0
Total Conventionally Indirect	32	60	11	73	9	56	11	61	3	75
Non-Conventionally Indirect
Hint	1	2	0	0	2	12	1	5	0	0
Total Non-Conventionally Indirect	1	2	0	0	2	12	1	5	0	0
Unclear/Incomplete Requests	0	0	1	6	0	0	0	0	0	0
Total Head Acts	53	100	15	100	16	100	18	100	4	100

. Data for this role play was not available for one NNS participant at the B2 level.

As shown in the previous table, both NSs and NNSs demonstrated a similar use of requests, with a strong preference for CI Requests. Specifically, the Query Preparatory of Permission (n = 17) was highly utilized not only by the NSs, but also by the NNSs, who exhibited a clear preference for this request strategy (A2: n = 8; B1: n = 6: B2: n = 8). Additionally, NSs also seemed to strongly favor Direct Requests in this scenario, such as the Need/Want Statement and the Query Preparatory of Permission. In contrast, NNSs relied less on Direct Requests overall. In order to illustrate the use of CI Requests in the Deadline Extension Scenario, examples for all levels are provided below:

(5)	a.	A2. Μπορώ να έχω *πολύ χρόνο; [Query Preparatory of Permission]
		Eng. Can I have more time?
	b.	B1. Μπορώ να σας δίνω την επόμενη εβδομάδα; [Query Preparatory of Permission]
		Eng. Can I give (it) to you next week?
	c.	B2. Μπορώ να σας στείλω την έκθεση λίγο αργότερα; [Query Preparatory of Permission]
		Eng. Can I send you the report a bit later?
	d.	C1. Ήθελα να ρωτήσω αν μπορείτε να μου δώσετε λίγο πιο χρόνο ακόμα για να το τελειώσω. [Query Preparatory of Ability]
		Eng. I wanted to ask you if you could give me a bit more time to finish it.
	e.	NS. Ίσως θα μπορούσα να ζητήσω κάποιες λίγες μέρες παραπάνω προθεσμία κατ’ εξαίρεση; [Query Preparatory of Permission]
		Eng. Maybe I could ask for a few days as an exception?

The means and standard deviations of the use of the three main types of requests across groups are provided in

Table 16. Means and standard deviations of the use of requests by NSs and NNSs in the Deadline Extension Scenario.

Groups		Direct	Conventionally Indirect	Non-Conventionally Indirect
NS	Mean	5.00	8.00	1.00
	Standard Deviation	6.96	7.18	0
A2	Mean	0.75	2.75	0
	Standard Deviation	1.30	3.11	0
B1	Mean	1.25	2.25	2.00
	Standard Deviation	2.16	2.27	0
B2	Mean	1.50	2.75	1.00
	Standard Deviation	2.60	3.27	0
C1	Mean	0.25	0.75	0
	Standard Deviation	0.43	0.83	0

.

The nominal logistic regression analysis failed to yield reliable statistical results for the types of head acts used by NSs and NNSs in this scenario. Consequently, the data for this scenario could not be included in the inferential analysis, as it did not meet the assumptions required for reliable statistical modeling.

3.3.2. Modifications

Figure 6 below presents the frequency of internal and external modification use across the different groups in this scenario.

The previous figure highlights a stronger preference for external modifications across groups. After the NSs, the B2 group showed the greatest use of modifications in this role play compared to the other proficiency levels. Notably, it can be observed that the frequency of external and internal modifiers increased with proficiency from levels A2 to B2. Participants at the C1 level also showed frequent use of modifiers.

Table 17. Means and standard deviations of the use of modifications by NSs and NNSs in the Deadline Extension Scenario.

Groups		External	Internal	Total
NS	Min	1	0	1
	Max	9	3	11
	Mean	3.40	1.62	5.02
	Standard Deviation	1.60	0.76	1.84
A2	Min	1	0	1
	Max	5	2	5
	Mean	2.33	0.53	2.87
	Standard Deviation	1.11	0.74	1.30
B1	Min	0	0	0
	Max	4	2	6
	Mean	2.44	0.56	3.00
	Standard Deviation	1.26	0.81	1.41
B2	Min	0	0	1
	Max	5	2	6
	Mean	2.47	0.89	3.37
	Standard Deviation	1.30	0.73	1.67
C1	Min	2	1	4
	Max	4	3	6
	Mean	3.00	2.25	5.25
	Standard Deviation	0.81	0.95	0.95

provides the means and standard deviations for the use of external and internal modifications by group.

A one-way ANOVA revealed statistically significant differences between groups regarding the number of request modifications in the Deadline Extension Scenario (F(9, 188) = 4, p < 0.001). The effect size, as measured by eta squared (η² = 0.265), indicated a large effect. Post hoc tests using Tukey’s HSD revealed significant differences between the following groups: A2 and NSs (p < 0.001), B1 and NSs (p < 0.001), and B2 and NSs (p = 003). To illustrate this, examples from the Deadline Extension Scenario are provided for A2, B1, B2, and NSs, highlighting the differences across these groups:

(6)	a.	A2. Θα ήθελα να σου λέω κάτι. [Preparator] Δεν είχα καιρό αρκετά να τελειώσω το δουλειά [Grounder] […] χρειάζομαι λίγο [Understater] *πιος καιρό.
		Eng. I would like to tell you something [Preparator] I didn’t have enough time to finish the assignment [Grounder] […] I need a bit [Understater] more time.
	b.	B1. Έχω, έχω πολλή δουλειά στο πανεπιστήμιο και δεν μπορώ να τελειώσω τη εργασία για μέρα [Grounder]. Μπορώ να έχω πιο πολύ χρόνο; παρακαλώ [Politeness Marker].
		Eng. I have, I have a lot of work at university, and I can’t finish the assignment for today [Grounder]. Can I have more time? Please [Politeness Marker].
	c.	B2. Έχω ένα μικρό πρόβλημα με τα με την εργασία που μου πρέπει να σας δώσω [Preparator] και θα ήθελα [Conditional] να έχω λίγο [Understater] πιο χρόνο.
		Eng. I have a little problem with the assignment that I have to give you [Grounder], and I would like [Conditional] to have a bit [Understater] more time.
	d.	NS. H εργασία που μας είχατε βάλει για να γράψουμε μέχρι σήμερα [Discourse Orientation Move], δυστυχώς λόγω κάποιων υποχρεώσεων που είχα δεν κατάφερα να την ολοκληρώσω [Grounder]. Θα μπορούσατε [Conditional], σας παρακαλώ [Politeness Marker + Solidarity], να μου δώσετε μία βδομάδα περιθώριο να σας τη φέρω; Θα με εξυπηρετούσε πολύ αυτό [Grounder], αν μπορούσατε αυτό να το κάνουμε [Considerator].
		Eng. The assignment that you gave us to write today [Discourse Orientation Move], unfortunately, due to some obligations I had, I couldn’t finish it [Grounder]. Could you [Conditional], please [Politeness Marker + Solidarity], give me an extra week to bring it to you? It would really help me if we could do that [Considerator].

Table 18. Overall distribution of external modifications used by all groups in the Deadline Extension Scenario.

External Modifications		Deadline Extension
		NS	A2	B1	B2	C1
Attention Getters	f	19	4	2	2	0
	%	10	11	5	4	0
Preparators	f	27	4	8	8	2
	%	15	11	20	17	16
Getting a Precommitment	f	1	0	1	0	0
	%	−1	0	2	0	0
Grounder	f	84	20	18	23	8
	%	46	57	46	48	66
Disarmer	f	6	2	1	2	0
	%	3	5	2	4	0
Imposition Minimizer	f	0	0	2	3	0
	%	0	0	5	6	0
Promise of Reward/Future Action	f	1	1	2	0	0
	%	−1	3	5	0	0
Apology	f	4	1	0	2	0
	%	2	3	0	4	0
Considerator	f	16	0	2	0	1
	%	8	0	5	0	8
Discourse Orientation Move	f	19	2	3	7	1
	%	10	5	7	14	8
Sweeteners	f	3	1	0	0	0
	%	1	3	0	0	0
Total	f	62	9	20	20	9
	%	61	81	74	60	56

provides the distribution of external modifier types employed by NSs and all NNS groups in this role play.

As can be observed in the previous table, the most frequently employed external modifier across groups was the Grounder, with NSs favoring it the most (n = 53). Similarly, NNSs also relied on Grounders, but their usage followed a non-linear trend (A2: n = 15; B1: n = 13; B2: n = 15; C1: n = 4). Moreover, all the groups used Preparators, but to a lesser extent in this scenario; NSs exhibited a high frequency of use (n = 24) and NNSs participants also displayed notable use (B1: n = 8; B2: n = 7). As for other strategies favored by the NSs in this role play, such as Discourse Orientation Moves (n = 19), Attention Getters (n = 17), and Considerators (n = 14), NNSs used them minimally.

Regarding internal modification devices,

Table 19. Overall distribution of internal modifications used by all groups in the Deadline Extension Scenario.

Internal Modifications			Deadline Extension
			NS	A2	B1	B2	C1
Syntactic	Negation	f	0	0	0	0	0
		%	0	0	0	0	0
	Subjunctive	f	0	0	0	0	0
		%	0	0	0	0	0
	Conditional Structure	f	28	0	2	4	1
		%	32	0	22	23	11
	Past Tense	f	3	0	0	1	1
		%	3	0	0	6	11
	Future Tense	f	3	0	0	1	0
		%	3	0	0	6	0
	Aspect	f	2	0	2	0	1
		%	2	0	22	0	11
	Passive Voice	f	0	0	0	0	0
		%	0	0	0	0	0
Lexical/Phrasal	Understaters/Hedgers	f	20	4	2	5	3
		%	23	50	22	29	33
	Politeness Marker	f	4	1	2	1	0
		%	4	12	22	6	0
	Subjectivizers	f	0	1	0	0	0
		%	0	12	0	0	0
	Downtoners	f	9	1	0	3	1
		%	10	12	0	17	11
	Cajolers	f	0	0	0	0	0
		%	0	0	0	0	0
	Solidarity Markers	f	3	0	0	0	0
		%	3	0	0	0	0
	Consultative Devices	f	13	1	1	2	2
		%	15	12	11	11	22
	Appealers	f	0	0	0	0	0
		%	0	0	0	0	0
	Intensifiers (Upgraders)	f	1	0	0	0	0
		%	1	0	0	0	0
Total		f	86	8	9	17	9
		%	32	18	18	26	42

displays the frequency of various syntactic and lexical/phrasal modifiers used by each group.

From the table above, it can be observed that NSs frequently employed the Conditional Structure (n = 28), in addition to the Past Tense and Future Tense (both n = 3) but to a lesser degree. In contrast, the overall use of syntactic modifiers by NNSs was scarce, with only some participants at the B2 level relying on the Conditional (n = 4). Additionally, the table highlights the NSs’ preference for Understaters/Hedgers (n = 20), Consultative Devices (n = 13), and Downtoners (n = 9). Of these three types of lexical/phrasal modifications, the NNSs primarily opted for Understaters/Hedgers, though its usage followed a non-linear trend (A2: n = 4; B1: n = 2; B2: n = 5; C1: n = 3).

3.4. Results of Retrospective Verbal Reports

The retrospective reports provided insightful information about the participants’ use of requests, offering a general overview of the NNSs’ self-perceptions and pragmatic concerns across proficiency levels.

The Shoes Scenario (−P, +D, −I) was considered the most familiar, easiest, and most frequently practiced role play in class, particularly for learners at the A2, B2, and C1 levels, due to the basic vocabulary, simpler grammatical structures, and participants’ previous experiences (see Examples 7a, 7b). The Cleaning Scenario (−P, −D, +I) was reported by several A2 and B1 learners as being moderately familiar, due to personal experience living with other people in the same apartment (see Example 7c). At the C1 level, two participants considered this scenario to be difficult since it involves a socially sensitive situation where the request can sound awkward. Lastly, the Deadline Extension Scenario (+P, +D, +I) was generally familiar to participants across proficiency levels due to academic or professional experience, although several learners at the A2, B1, and C1 levels reported difficulty finding appropriate vocabulary and expressions due to formality (see Examples 7d, 7e).

(7)	a.	A2. Si, τα μαγαζιά muchas veces, cosas en clase, pocas y del resto poco, hacemos mucho de μαγαζιά, πελάτης…
		Eng. Yes, the shops many times, things in class, very few, and the rest very little, we do a lot about the shops, client…
	b.	B2. Στη ζωή μου, όταν εγώ ήμουν στην Ελλάδα, εγώ αγόρασα πράγματα και εγώ, έπαιξα τον ρόλο αυτό, στα μαγαζιά, και πιστεύω ότι εγώ ξέρω καλά τις λέξεις των ρούχων και εδώ στo μάθημα επίσης.
		Eng. In my life, when I went to Greece, I bought things and I played this role, in the shop, and I think that I know the words about clothes well and in the class too.
	c.	B1. Este contexto lo conozco (Cleaning) por la familiaridad y porque las palabras son más sencillas.
		Eng. I’m familiar with this context (Cleaning) because I know it well and the vocabulary is simpler.
	d.	A2. Quizás está la de… (Deadline Extension), me parece que en Grecia es señor profesor, aquí no.
		Eng. Perhaps this one of… (Deadline Extension), I think in Greece it is “Mr. Professor”, not here.
	e.	C1. Με αυτή (Deadline Extension), δεν ξέρω, για τις λέξεις, λόγω των λέξεων, δεν ξέρω, για μένα είναι δύσκολα όλα.
		Eng. With this (Deadline Extension), I don’t know, for the words, because of the words, I don’t know, for me all are difficult.

4. Discussion

The first research question aimed at exploring whether proficiency plays a role in the production of requests (head acts and modifications) in Greek as an FL by Spanish/Catalan bilinguals. The findings revealed that proficiency plays a nuanced role in the acquisition of Greek requests by these participants. Across scenarios, the data confirms a trend consistent with that in the literature, that although increased proficiency supports greater grammatical and pragmalinguistic complexity, sociopragmatic competence often lags behind (Bardovi-Harlig, 1999, 2001; Félix-Brasdefer, 2003; Bella, 2012b).

In the Cleaning Scenario (−P, −D, +I), clear proficiency-related differences were observed in the types of head acts used. While NSs predominantly relied on NCI Requests, NNSs used a range of forms, with A2, B1, and C1 learners favoring CI Requests and B2 learners displaying a strong preference for Direct Requests. Some A2 and B1 participants used NCI forms more frequently than expected, possibly as a compensatory strategy (Trosborg, 1995) due to limited grammatical resources needed for CI constructions (Bella, 2012b, 2014c). In contrast, B2 learners’ overuse of Direct Requests—deviating from native speaker norms—may reflect an overestimation of their right to be direct, revealing gaps in sociopragmatic awareness despite their advanced grammatical competence. B2 learners may have perceived a right to impose in this situation (Economidou-Kogetsidis, 2018; Kasper & Rose, 2002). In addition, this tendency may reflect their L1 cultural mindset (Kecskés, 2013), as B2 learners could have transferred culturally specific norms of directness into Greek, overgeneralizing their sense of entitlement to make Direct Requests and failing to adjust to the social parameters of the target language appropriately. C1 learners, on the other hand, displayed more target-like forms by prioritizing politeness through CI Requests, although this result should be interpreted cautiously due to the limited size of the C1 group.

In the Shoes Scenario (−P, +D, −I), proficiency did not produce a clear developmental pattern in head act usage. Across all proficiency levels, NNSs overwhelmingly relied on Direct Requests, contrasting with NSs, who predominantly used CI forms. This overuse of directness, possibly influenced by their L1 cultural mindset (Kecskés, 2013), was statistically significant—particularly for A2 and B2 learners—and it points to incomplete pragmatic development, even at more advanced stages. It also highlights the need to distinguish linguistic from sociopragmatic competence (Bardovi-Harlig & Dörnyei, 1998; Taguchi, 2011).

In terms of modifications, significant differences emerged in both the Shoes and Deadline Extension scenarios. In the Shoes Scenario, A2 learners used fewer and less varied modifications compared to C1 and NS participants, relying heavily on formulaic or simple structures (Kasper & Rose, 2002). Similarly, in the Deadline Extension Scenario (+P, +D, +I), A2 participants relied predominantly on basic modifiers and formulaic expressions like “please” (παρακαλώ) and “I would like” (θα ήθελα). While B1 and B2 learners showed modest progression in the range of modifications used, they still fell short of NS norms in terms of variety and frequency, reflecting delayed sociopragmatic development (Bardovi-Harlig, 1999, 2001; Barron, 2003). Learners at the B1 and B2 levels particularly struggled with the Deadline Extension scenario, as revealed in the retrospective verbal reports, mentioning unfamiliarity, lack of resources, or discomfort when making polite requests in unequal-status settings. In contrast, in the Cleaning Scenario, no statistically significant differences were observed across groups in modification use. This may reflect learners’ shared perception of the situation as more urgent and less face-threatening, which possibly minimized the need to employ mitigation strategies (Ackermann, 2023; Brown & Levinson, 1987). Across all three scenarios, learners consistently favored external over internal modifiers (as in Bella, 2012b), likely due to the relative ease with which they can be produced. External modifiers involve simpler syntax and are less demanding than internal ones, both linguistically and cognitively (Economidou-Kogetsidis, 2009). Again, learners may have relied on external modifiers as a compensatory strategy (Trosborg, 1995) to offset their limited use of internal modifiers.

Turning to the second research question, the study examined the role of sociopragmatic variables (Power, Social Distance, and Imposition) on the production of requests in Greek as an FL by Spanish/Catalan bilinguals across different proficiency levels. The findings revealed that the sociopragmatic variables play a significant role in shaping request behavior among Spanish/Catalan bilingual learners of Greek as an FL. However, their impact varies across proficiency levels and contexts.

In the Cleaning Scenario (−P, −D, +I), despite low Social Distance and Power, the high level of Imposition made the situation particularly face-threatening. NSs mitigated this through a preference for NCI and CI Requests, aligning with Greek politeness norms emphasizing involvement and face protection (Economidou-Kogetsidis, 2003; Sifianou, 1992a, 1999). NNS participants, however, deviated from this norm, especially at the B2 level, where Direct Requests were dominant. This suggests that learners may misinterpret the social dynamics in high-Imposition but low-Power/Distance scenarios, failing to align formality with the degree of face threat.

In the Shoes Scenario (−P, +D, −I), the influence of Social Distance was evident. NSs maintained politeness by favoring CI Requests, while NNS learners across all levels used Direct Requests. This may reflect a perception of the context as a routine, low-Imposition service exchange (Economidou-Kogetsidis, 2005), where learners felt entitled to request directly. This interpretation is supported by Gkouma (2024) and the verbal reports, which might indicate that many learners viewed the scenario as familiar and routine. The scenario’s everyday nature, frequent classroom practice, and vocabulary simplicity likely reinforced the perception that Direct Requests were socially acceptable, especially given the influence of L1 norms from Spanish (Escandell-Vidal, 2005; Pinto, 2012; Ruiz, 2018) and Catalan (Curell, 2012). However, this assumption resulted in pragmatic failure, as NS norms in Greek still favor indirectness, even in transactional exchanges.

In the Deadline Extension Scenario (+P, +D, +I), both NS and NNS participants showed a shared preference for CI Requests, specifically the Query Preparatory of Permission. This scenario elicited more appropriate sociopragmatic behavior from NNSs, who recognized the need for deference and mitigation. This contrasts with Bella’s (2012b) findings, where such request forms were generally absent among lower–intermediate and intermediate learners in the same type of scenario. The similarity in the use of CI Requests in the Deadline Extension scenario may be due in part to cross-cultural pragmalinguistic similarities in Spanish (Pérez-Ávila, 2005) and Catalan requests (Vanrell & Catany, 2021). Learners’ familiarity (Gkouma, 2024) with academic scenarios involving request extensions, as noted in the verbal reports, also likely contributed to this appropriate performance. Modification use, however, remained limited, particularly among A2 and B1 learners. Yet all groups—including NSs—relied heavily on Grounders in the Deadline Extension Scenario, indicating a shared understanding of the need for justification in formal, high-Imposition contexts. This supports previous findings that Grounders are a key modification strategy in formal requests (Bella, 2012b; Economidou-Kogetsidis, 2011; Félix-Brasdefer, 2007; Schauer, 2007).

In conclusion, the findings of the present study support the perspective that increased proficiency does not always align with more target-like pragmatic performance, as documented by several studies (e.g., Barron, 2003; Bardovi-Harlig, 1999; Taguchi, 2011). This discrepancy highlights the need for targeted instructional approaches. Accordingly, several pedagogical implications can be drawn, emphasizing the importance of explicitly integrating sociopragmatic instruction into Greek as an FL curricula and addressing not only grammatical competence but also pragmatic awareness in language teaching. Additionally, it was observed that while some learners demonstrated sensitivity to social parameters, especially in high-Imposition or formal scenarios, likely due to increased perceived risk and familiarity, others did not consistently adjust their request strategies accordingly, which may imply limited pragmatic development (Achiba, 2003; Schauer, 2004; Woodfield & Economidou-Kogetsidis, 2010). Therefore, pragmatic instruction should be context-sensitive, helping learners interpret and respond to variable social dynamics appropriately. Lastly, variations in request use might also reflect learners’ subjectivity or agency (Ishihara, 2019). That is, learners in the present study may have made intentional pragmatic choices (even when these diverge from perceived target norms) shaped by their identities, values, and communicative goals. In such cases, variability in request use cannot be attributed solely to L1 transfer or limited awareness of L2 norms. Rather, learners may have exercised agency by negotiating a personal pragmatic space in which target norms are balanced with their own linguistic priorities and their L1 cultural mindsets. In this sense, instruction should embrace learners’ diverse linguistic and cultural identities while fostering adaptability to varying communicative contexts, rather than focusing solely on target-like forms.

The present study acknowledges several limitations, which deserve attention and offer valuable directions for future research. First, despite a relatively large overall sample, uneven distribution across proficiency levels (especially considering that the study included only four participants at the C1 level) may have limited statistical power, thereby reducing the generalizability of the findings. It should also be noted that the present study is cross-sectional. To gain a deeper understanding of pragmatic development in the use of requests, future research should consider conducting longitudinal studies to track L2 learners’ pragmatic development over time, as suggested by other researchers (e.g., Taguchi, 2018; Timpe-Laughlin, 2017).