Contextual Variables as Predictors of Verb Form: An Analysis of Gender and Stance in Peninsular Spanish Requests

Abstract

The current experiment employs a variational pragmatics perspective to explore how the contextual variables of power, distance, and imposition jointly affect social groups’ and individuals’ choice of verb forms in requests in Madrid, Spain. Using a mixed-method approach to explore the requests of 111 Spanish speakers from Madrid, quantitative analyses determined the level of significance and hierarchical order of the predictor variables of power, distance, and imposition on verb form and also the distribution of verb forms by gender, with male and female participants exhibiting significant differences. Additionally, certain participants demonstrated decreased sensitivity to contextual factors, adopting more categorically indirect or direct request strategies. The examination of both gendered request patterns and the stances that single participants adopt through their verb-form selections contributes to our understanding of the social moves that are made by all speakers, not just those who fall within the gendered norms. The results highlight the different frames and social meanings attached to these forms at the micro- and macro-social levels, providing new insight into the complex relationship among linguistic variables, contextual factors, and social groups and individuals.

1. Introduction

Linguistic forms used in social interaction reflect and reconstruct identities, relationships, situations, and cultures. Whereas sociolinguistics has highlighted the social indexing of linguistic forms (Agha 2007; California Style Collective 1993; Eckert 2016), the field of variational pragmatics provides a complementary approach that addresses the relationships among social factors and language use in context (Acton 2021; Barron 2014). The field examines how particular uses of linguistic forms, in particular contexts and conversations, reflect and construct the social world.1 When defining interactional contexts, previous research in pragmatics and politeness has often considered the role of the key contextual factors of power (e.g., peer vs. superior), distance (e.g., known vs. stranger), and imposition (e.g., small vs. big favor) (e.g., Brown and Levinson 1987; Félix-Brasdefer 2005, 2009; Hübscher et al. 2017), but analyses have seldom systematically examined the joint impact of contextual factors or the interrelationship among them. Therefore, questions remain about how the contextual variables of power, distance, and imposition jointly affect language use, how they shape the language choices of different speakers, and what those findings indicate about social meaning.

This investigation examines how contextual factors (i.e., power, distance, and imposition) condition verb forms used by men and women in Madrid, Spain; the variation in verb-form use by men and women; and how Peninsular Spanish verb forms contribute to individual speakers’ stances when making requests in contexts that vary in power, distance, and imposition. The theoretical frame employed recognizes the connection between contextual variables and linguistic production as well as the relationship between individuals’ stances and macro-level sociolinguistic trends, aligning with variational pragmatics (Acton 2021; Barron 2014), third-wave sociolinguistic research (e.g., Bell 2016; Eckert 2012, 2016; Eckert and McConnell-Ginet 1999), and Terkourafi’s (2005, 2015) frame-based approach to politeness.

The contribution of this paper is threefold. First, as one of the few studies to quantitatively investigate the effect of the interaction of power, distance, and imposition on linguistic forms, it demonstrates that contextual variables form a specific hierarchical relationship that can be conditioned by social group. Thus, the topic of variation is addressed, but unlike most prior linguistics research, this investigation highlights the variable perception of context and its effect on language. Second, this investigation offers new insight into language and gender in the Peninsular Spanish context by examining differences in the effect of contextual variables on men’s and women’s language, as well as differences in verb-form selection by men and women from Madrid. Third, the analyses address individual-level data that demonstrate the agency of individuals in selecting verb forms that reflect their preferred stances in particular interactional contexts. The results identify nuanced social meanings at group and individual levels that emerge from the study of contextual variables, linguistic forms, and individual stances as dynamic aspects of interaction.

2. Literature Review

2.1. Pragmatic Variation and the Social Indexing of Language

Pragmatics research has often addressed linguistic variation across situations but paid little attention to how situational variation intersects with social variation. With the distinction of the field of variational pragmatics, which initially intended to integrate pragmatics and dialectology (Barron 2005), more researchers have begun to consider the relationship between pragmatic variables and social factors such as social class, ethnicity, age, education, religion, and gender (Barron and Schneider 2009). Demonstrating the need for a pragmatic approach to social variation, D’Arcy’s (2017) research highlighted the multiple functions of like in interaction, which do not all behave similarly in their social uses and distributions. Her analyses countered the notion that women employ like more than men by showing that different pragmatic uses are favored by men and women. In general, the variational pragmatics approach examines how particular uses of linguistic forms in particular contexts and conversations reflect and construct the social world. This highlights the context-dependency of language that influences language variation and reflects the individual identities of language users (Acton 2021). Additionally, variational pragmatics foregrounds the questions addressed in this article about the varying effects of situations on different speakers, as evident in their linguistic production.

The theoretical frame of this investigation blends concepts and understandings from pragmatics and variationism. As Acton (2021) argues, pragmatics and third-wave variationism share many underlying principles and provide complementary approaches to understanding social meaning. In addition to adopting a pragmatics perspective of language and context, the current research recognizes the indexical relationship between linguistic variables (e.g., verb forms) and social factors (e.g., gender), a relationship that serves to (re)create social structures. Indexicality (Agha 2003) and Silverstein’s (2003) concept of indexical order expose the relationship between local uses of linguistic features in interaction and their link to styles, performances, and macro-social identities (e.g., California Style Collective 1993; Jaffe 2016; Silverstein 2003).

Concerning specifically the relationship between linguistic forms and macro-social identity categories, stance is an important mediator (Ochs 1996). It has been shown that ‘sociolinguistic variants are initially associated with interactional stances and these stances become in turn associated with a social group meaning’ (Kiesling 2009, p. 172) (see also Davies and Harré 1990; Du Bois 2007; Eckert 2016; Jaffe 2016). As defined by Du Bois (2007), ‘[s]tance is a public act by a social actor, achieved dialogically through overt communicative means, of simultaneously evaluating objects, positioning subjects (self and others), and aligning with other subjects, with respect to any salient dimension of the sociocultural field’ (p. 163). As an example of stance being a mediator between linguistic variables and social group identity, Kiesling (2004) argues that the use of the address term dude is first associated with a stance of ‘cool solidarity,’ which then is associated with masculinity (i.e., a gendered social identity) (p. 282). He states that ‘if a linguistic item co-occurs frequently in the speech of a particular person or kind of person, that linguistic item will be taken to index that group’ (Kiesling 2009, p. 177).2 In this paper, the analysis of stance is limited to how verb forms position speakers and hearers with respect to each other and, also, with respect to the request being made in the interaction.

2.2. Politeness: Linguistic Expressions, Speaker, and Context

One area of study within pragmatics that has focused on social meaning is that of politeness. When being polite, the purpose is to maintain relationships and facilitate interactions by behaving in ways that are expected by other interlocutors (e.g., Kádár 2019; Spencer-Oatey 2000; Watts 2003; Terkourafi 2015). To be polite, speakers use conventionalized language that is processed as polite via generalized conversational implicatures (Terkourafi 2005, 2015).3 Conventionalized linguistic forms are socially constituted via the three-way relationship among ‘an expression, a context, and a speaker’ (Terkourafi 2015, p. 15). Whereas sociolinguistic theories have explained these connections via stance, indexicality, and situations (e.g., Ochs 1996), Terkourafi (2005, 2015) relies on the concept of a frame, which is a cognitive construct based on prior experience that includes linguistic and non-linguistic information and that can be recalled from memory to aid in interaction (e.g., Ensink and Sauer 2003; Fillmore 1975; Schank and Abelson 1975). For Terkourafi (2015), frames are combinations of an expression with a minimal context that includes extra-linguistic variables such as gender, age, relationship, setting, etc. Speakers’ frames provide predictive capabilities about language use in particular contexts, and they reflect individual and social group experiences, meaning that frames may vary across speakers and time (Terkourafi 2005, p. 15). Terkourafi’s (2005, 2015) approach to politeness aligns with the social constructivist theories in Section 2.1 in that any aspect of the frame can be understood as socially constituted. It also offers the theoretical explanation of a cognitive frame as a resource in interaction.

Section 2.2.1, Section 2.2.2 ans Section 2.2.3 introduce aspects of the three-way relationship among the linguistic expression, speaker, and context that are relevant to the current investigation. Section 2.2.1 addresses verb forms in requests, Section 2.2.2 provides prior research related to the extra-linguistic variable of gender, and Section 2.2.3 introduces the contextual variables that are explored in this study.

2.2.1. Linguistic Expression: Requests and Verb Forms

Requests are speech acts that attempt to make a hearer do something (Austin [1962] 1975; Searle 1976). Requests can be realized with direct, conventionally indirect, or non-conventionally indirect linguistic strategies (Blum-Kulka 1987; Brown and Levinson 1987). The requests that are most direct in languages like Spanish are accomplished with imperative verbs (Blum-Kulka et al. 1989), and the imperative is more commonly found in Spanish when the speaker has more social power compared to the interlocutor (Hernández-Flores 2004). While the imperative is not uncommon in Peninsular Spanish (Ballesteros 2001; Lorenzo-Dus and Bou-Franch 2003), conventionalized indirect requests (e.g., ¿Puedes pasarme la sal? ‘Can you pass me the salt?; ¿Podrías pasarme la sal? ‘Could you pass me the salt?’) seem to be the most frequent type (Márquez-Reiter 2002, 2003; Márquez-Reiter et al. 2005). In fact, the frequent use of conventionalized indirect requests is common across the Spanish-speaking world, with evidence from Uruguayan Spanish (Márquez-Reiter 2000, 2002, 2003), Mexican Spanish (Félix-Brasdefer 2005, 2009), and Costa Rican Spanish (Félix-Brasdefer 2009). While the directness or indirectness of a request is important, the verb form is a main indicator of the illocutionary force (Félix-Brasdefer 2005), and it communicates detailed meaning beyond the direct/indirect strategy distinction.

The verbal system in Spanish, with a variety of tenses and moods, provides rich semantic options for making requests and also nuanced pragmatic meanings. In general, the deictic center of the verb is commonly used to convey different degrees of politeness in requests (Koike 1989). For example, the present tense communicates a more-direct illocutionary force, whereas future tense or conditional verb forms are less imposing (Koike 1989). Conditional, past, and future forms are temporally displaced from the current moment and thus present a temporal disorder that is linked to the expression of politeness (Fleischman 1989; Koike 1989). Furthermore, verb forms such as imperfect and future forms are representative of an irrealis mood, which has implications for politeness and can ‘modulate the perceived assertiveness’ (Fleischman 1989, p. 8). Not only are irrealis or non-factual forms associated with more politeness due to their temporal distance from the action as compared to realis or factual forms, but politeness becomes encoded in the verb forms (Chodorowska-Pilch 1998, 2004).

Research on verb forms in varying contexts has highlighted the connections among verb forms, mitigation, and politeness. In a study on Mexican Spanish, conditional and imperfect verb forms were used as syntactic mitigators in situations that displayed a distant relationship between interlocutors (Félix-Brasdefer 2009). Modal verbs such as poder ‘to be able to’ (e.g., Can you pass me the salt? vs. Could you pass me the salt?) were also found as mitigators when addressing a distant person, expressing a higher level of deference when used in the conditional or imperfect forms compared to present tense (Félix-Brasdefer 2005). Modal verbs mitigate by suspending to some degree the assumption that the hearer is able and willing to perform the requested action (Brown and Levinson 1987; Briz Gómez 2004, pp. 72–76)—a concept that refers to the preparatory condition of directives (Searle 1976) and relates to deictic distancing.

Based on the prior literature, and following primarily Chodorowska-Pilch (1998, 2004), the conventional politeness of verb forms in Spanish can be understood as a continuum: subjunctive, modal conditional, conditional, imperfect/future, modal present, present, imperative (Figure 1).4 The high end of the continuum represents verb forms with an irrealis mood and a deictic distance from the action. The low end of the continuum represents verb forms with a realis mood that maintain the present deictic reference. The forms closest to the high end have been described as less direct and less assertive; those near the low end have been described as more direct and more assertive (Koike 1989; Fleischman 1989). All verb forms in Figure 1 are commonly used by speakers of Spanish. Excerpts (1)–(8) from the current data set demonstrate the use of each verb form in a request.5 As can be noted in the excerpts, a multitude of other linguistic resources are used in conjunction with verbs to make requests. While the co-occurrence of linguistic resources can be important in communicating social meaning (e.g., California Style Collective 1993; Eckert 2012), the current focus is on the verb because the verb is a main indicator of the illocutionary force (Félix-Brasdefer 2005) and the directness or assertiveness of a given utterance (Koike 1989; Fleischman 1989).

(1)

Subjunctive.

Perdone, necesito un manual de referencia para realizar mi trabajo, si pudiese dejármelo únicamente durante este fin de semana se lo agradecería.

Excuse me, I need a reference manual to do my project, if you could lend it to me just for this weekend I would appreciate it.

(2)

Modal Conditional.

Mirian ¿me podrías dejar una grapadora para el trabajo, por favor?

Mirian, could you lend me a stapler for the project, please?

(3)

Conditional.

Perdone, ¿tendría todavía alguna hoja de las que entregó el último día?

Excuse me, would you still have any of the sheets you handed out the last day?

(4)

Imperfect.

Hola Rosali, oye, ¿cuál era el autor del artículo del otro día?

Hi Rosali, listen, which was the author of the article the other day?

(5)

Future.

Rosa, ¿no tendrás un boli de sobra? que no encuentro el mío.

Rosa, do you happen to have a spare pen? It’s that I can’t find mine.

(6)

Modal Present.

Luis, ¿me puedes ayudar con el examen?

Luis, can you help me with the exam?

(7)

Present.

Sergio, ¿me das un folio?

Sergio, will you give me a sheet of paper?

(8)

Imperative.

Déjame un folio, tío.

Give me a sheet of paper, man.

The verbal continuum aligns with the notion that through continual societal sanctioning of the use of certain forms in certain situations, a form becomes conventionalized (Chodorowska-Pilch 1998; Terkourafi 2005, 2015). Thus, certain forms come to index a degree of politeness (Agha 2007; Pizziconi 2011). Whereas the described pragmatic meanings of verb forms within a spectrum of directness emerge from a traditional pragmatics perspective in which these systems of meaning are available for use by all speakers, the current investigation aims to provide nuance to the question of who uses these forms and under what circumstances.

2.2.2. Speaker: The Variable of Gender

Gender, like other social identities, affects and reflects one’s social experience and one’s use of linguistic resources. From a performance-based perspective, identities are malleable, and they are socially constructed via language and interaction through ‘the social positioning of the self and other’ (e.g., Bucholtz and Hall 2005, p. 586; Gee 2014). In practice, one can demonstrate or create group membership through linguistic resources that are associated with that group; this relationship to group membership is not fixed but rather it varies across time, context, conversational topic, etc. (e.g., California Style Collective 1993; Eckert and McConnell-Ginet 1992).

The language of different gendered groups has been of interest for at least the last 50 years in linguistics research. Research on language and gender has alluded to and demonstrated in some cases correlations between gender and certain linguistic characteristics.6 Lakoff (1975) proposed that women rely more on tentative language (e.g., hedges, qualifiers, disclaimers, tag questions, intensifiers), which ‘softens an assertion’ (Leaper and Robnett 2011, p. 137). Confirming these claims, Leaper and Robnett (2011) conducted a metastudy that included 29 prior investigations of tentative language. They found a small but significant difference between men and women in the use of tag questions, markers of uncertainty, and intensifiers. The effect size was larger for longer interactions (vs. shorter), undergraduates (vs. other adults), group conversations (vs. dyads), and research settings (vs. others). The different effect sizes indicate that the construction of gendered identities may be more important during certain times of life (e.g., university) or in certain types of conversations (e.g., group conversations), demonstrating the malleability of one’s linguistically marked belonging to a gendered group.

The prior findings have been interpreted as functioning to facilitate interaction, express interpersonal sensitivity, and create a supportive and affective identity (Grenoble 1999; Hancock and Rubin 2015; Holmes 1995; Leaper and Robnett 2011; Levey 2003; Wright and Hosman 1983). However, Eckert (1989) and Uchida (1992) argued that power is the underlying sociological construct associated with gender, resulting from historical and current norms of limited participation in society for certain gendered identities, including those of women. This argument aligns with Lakoff’s (1975) explanation that women’s tentative language may be associated with their subordinate position in society. We draw attention to the previous findings about tentative language and gender because, similar to the linguistic resources analyzed as tentative language, verb forms with a deictic distancing from the action also serve a softening function (see Section 2.2.1) and, thus, may be differentially associated with gendered performances. At the same time, the current analyses address macro-level trends and individual tendencies considering that ‘gender does not have a uniform effect on linguistic behavior for the community as a whole, across variables, or for that matter for any individual’ (Eckert 1989, p. 253).

In the Spanish-speaking world, the relationship between gender and language has been understudied, and few investigations of speech acts have considered gender. In an analysis of request strategies and gender in Argentine Spanish, no differences in the use of direct questions, assertions, imperative use, or conventional indirect requests were found between men and women (Yates 2015). In another study, Lorenzo-Dus and Bou-Franch (2003) found that across various speech acts, Spanish women used alerters7 and acts of thanking more often than men, while using fewer supportive moves (utterances that provide additional explanation, justification, or support for a given speech act). In requests specifically, they found a preference for direct requests by both Spanish women and men, indicating that ‘cultural behavior may be a stronger factor than gender in this particular aspect of the formulation of requests’ (p. 9). They also found that women used more-direct strategies (i.e., want statements) than men in situations with asymmetrical power and distance (+power, +distance), where 30.76% of the women’s responses and 15.78% of the men’s were direct request strategies. These results provide mixed findings on the type of linguistic resources preferred by women versus men when performing speech acts in Spanish, indicating a need to further explore the relationship between language and gender in speech acts in the Peninsular Spanish context and in general.

2.2.3. Contextual Variables

The meaning of language is context-dependent. While various aspects of context are relevant to understanding language and social meaning, politeness research has confirmed the importance of three widely examined contextual variables: power, distance, and imposition (e.g., Brown and Gilman [1960] 1972; Brown and Levinson 1987; Brown et al. 2014; Czerwionka 2012, 2014; Félix-Brasdefer 2005, 2008, 2009; Márquez-Reiter 2002, 2003; Márquez-Reiter et al. 2005). According to Brown and Levinson (1987), power represents asymmetrical hierarchies that can be found in a society, distance is a symmetrical relationship that ranges from close to distant, and the situational variable of imposition is the degree to which an act is considered to interfere with the hearer’s wants or desires to be self-determined (Brown and Levinson 1987). Each of these contextual variables should be understood to encompass a range of interpretations. For example, Spencer-Oatey (1996) reported that distance has been described using terms such as solidarity, familiarity, closeness, and relational intimacy. Thus, the meaning of power, distance, imposition, or any other contextual variable must be scrutinized.

From a theoretical perspective, it is also important to note that the interpretations of power, distance, and imposition are socially constructed and may vary among different cultures or groups of speakers (Brown and Levinson 1987). For example, with respect to power, one can imagine distinct power differentials in an employee–boss relationship depending on the cultural, societal, or individual interpretations. With respect to imposition, particular cultures or individuals may understand the weight of a specific request or favor differently, with the same request being perceived as high imposition by some and low imposition by others. Referring to the variability in the interpretation of contextual factors and justifying the current research agenda, Spencer-Oatey and Žegarac (2017) indicated that ‘little (im)politeness research has grappled with such issues’ (p. 134).

While the impact or understanding of these contextual factors is variable, so too are the relationships among them. That is, the relative importance of these variables may change at different moments or from one community to another—a finding that has recently begun to emerge in empirical pragmatic research. Tamaoka et al. (2010), for example, examined the hierarchical organization of power, distance, and gender on perceived politeness in Japanese and Korean speakers. Using a decision tree analysis, the results demonstrated that Japanese participants’ responses were predicted by the distance between the interlocutors, followed by the power relationship. In contrast, the Korean participants’ responses were influenced by power differentials, while distance was not a significant predictor. The difference across groups suggests that the way in which Japanese and Korean speakers order the importance of these contextual variables in interactions is distinct, with Japanese speakers being most influenced by distance and Korean speakers most influenced by power.

The investigation by Tamaoka et al. (2010) supports the notion that contextual variables do not necessarily have the same social meaning across societies. The study also demonstrates that predicted hierarchical models can provide insight into how contextual factors shape behaviors and how they differentially constrain the language use of members of different social groups. Further investigation of this underexplored topic can provide insight into the socially determined understanding of the relationships among context, linguistic behaviors, and the social meaning of language.

2.3. Research Questions

This study explores the effect of power, distance, and imposition on language in Spanish, focusing on verb forms and considering the variable of gender. The primary goal is to examine how contextual factors condition the linguistic forms used by women and men in Madrid, Spain. We also examine how verb selection varies by gender and how individual-level linguistic choices portray different stances. The following research questions guide the analyses:

• Do the effect and hierarchical order of the contextual variables of power, distance, and imposition as predictor variables of verb-form selection in requests differ between women and men?
• Does the overall distribution of verb forms used in requests differ between women and men?
• How do individuals’ stances in requests vary in contexts that involve differing degrees of power, distance, and imposition?

Based on prior research, we hypothesize that power and distance will have a greater effect than imposition on verb selection and a higher placement in terms of the hierarchical structure (e.g., Ballesteros 2002; Félix-Brasdefer 2005, 2009; Le Pair 1996). No specific hypothesis was made about the hierarchical order across genders because of the lack of related prior research. For the second research question, there are two competing hypotheses. The first is that women will use less-direct verb forms, aligning with the tendency for women to use more tentative linguistic forms (e.g., Leaper and Robnett 2011). The second is that women will use more-direct verb forms, aligning with Lorenzo-Dus and Bou-Franch’s (2003) finding that women in Spain were more likely to use direct speech acts, at least in certain contexts. Considering the third research question, we expect that variation exists, confirming that speakers have the agency to select linguistic forms that serve to position them in ways that align with their ideologies, expectations, and identities (e.g., Du Bois 2007; Eckert 2016; Jaffe 2016).

3. Materials and Methods

This investigation relied on data collected using a discourse completion task (DCT) to prompt the production of requests in contexts with two levels of power, distance, and imposition.8 Quantitative and qualitative analyses were conducted to understand the effect of the contextual variables of power, distance, and imposition on verb selection in requests produced by men and women in Madrid, Spain; verb selection across genders; and individuals’ stances communicated with verb forms.

3.1. Participants

A total of 111 native speakers of Peninsular Spanish from the Madrid region of Spain (i.e., Community of Madrid) participated in the study after providing their consent (70 women and 41 men; Age: M = 26.6, SD = 6.60). Gender was elicited using a multiple-choice question with the options of male, female, or other, where the option of other included a fill-in-the-blank box. All participants had graduated from or were enrolled in a Spanish university degree program. The participants likely spoke the same or very similar varieties of Spanish, given the homogeneity of their geographic location and educational background. Additionally, the university-level educational experience of all participants was ideal for the current research because it ensured familiarity with the situations employed in the experiment. All reported participant names are pseudonyms.

3.2. Materials and Procedure

A DCT is an elicitation procedure where participants read a contextualized prompt and provide a response to it. DCTs have been widely used for pragmatics research because they allow for the collection of large amounts of data in controlled situations. They have also been criticized as they provide metapragmatic data rather than naturally occurring data (Golato 2003). While DCT data do not represent all aspects of a naturally occurring interaction (Félix-Brasdefer 2007; Golato 2003), DCT and naturally occurring data share more similarities when considering head acts (i.e., the part that communicates the speech act) than other aspects of the interactions (see Márquez-Reiter and Placencia’s (2005) discussion, p. 226). For requests, Bataller and Shively (2011) found similarities across data types in request openings and a similar variability in request type. Given the current focus on request head acts and the need for a large amount of controlled data for statistical analysis, DCT data were ideal.

The DCT used in the current investigation included 16 situations with controlled levels of power, distance, and imposition (Situation 11,

Table 1. Distribution of the situations.

Situation	Power	Distance	Imposition	Type of Request
1	Asymmetrical	Distant	High	Service
2	Asymmetrical	Distant	Low	Service
3	Asymmetrical	Distant	High	Good
4	Asymmetrical	Distant	Low	Good
5	Asymmetrical	Close	High	Service
6	Asymmetrical	Close	Low	Service
7	Asymmetrical	Close	High	Good
8	Asymmetrical	Close	Low	Good
9	Symmetrical	Distant	High	Service
10	Symmetrical	Distant	Low	Service
11	Symmetrical	Distant	High	Good
12	Symmetrical	Distant	Low	Good
13	Symmetrical	Close	High	Service
14	Symmetrical	Close	Low	Service
15	Symmetrical	Close	High	Good
16	Symmetrical	Close	Low	Good

, Appendix A). All were set in an academic environment in which a student interacted with another student (−power) who was a well-known friend (−distance) or a not-very-well-known classmate (+distance) and in which a student interacted with a professor (+power) who was either well known (−distance) or not very well known to the student (+distance). The situations required that a student request something of the other interlocutor; half of the situations prompted a low-imposition request (−imposition) and half prompted a high-imposition request (+imposition). The 16 situations were balanced considering two levels of each variable: power (asymmetrical [+power], symmetrical [−power]), distance (distant [+distance], closeness [−distance]), and imposition (high [+imposition], low [−imposition]). Half of the situations were about requesting a good and half were about requesting a service (i.e., type of request (Brown and Levinson 1987)), although the effect of the type of request was not of interest in the current study. The order of the 16 situations was computer-randomized.

Situation 11. −power, +distance, and +imposition.

Tienes un examen la próxima semana y has faltado a algunas clases. [Nombre de un estudiante al que no conoces mucho] se encuentra a tu lado. Quieres que te preste sus apuntes. ¿Qué le dices?

You have an exam next week and you have missed some classes. [Name of a student that you don’t know very well] is next to you. You want the student to lend you her/his notes for the exam. What do you say?

During the task, participants were first asked to provide the names of two professors and two classmates, one who was well-known and another who was not. These names were automatically embedded in the preambles seen by the individual participant to signal the social power and distance between interlocutors. Referring to the names of classmates and professors whom participants knew resulted in greater authenticity. Then, participants read each situation and wrote what they would say in response. Written responses facilitated the data analysis, and they were sufficient for the purposes of the current study. Following the DCT, participants also completed a background questionnaire that requested information about gender, age, place of origin (city), native language, and other spoken languages. All tasks were conducted online using Qualtrics (2018).

In the creation phase of the DCT, a norming procedure was conducted to assess the perception of the contextual variables. Twelve native speakers of Spanish, who were university students in Madrid and who were not participants in the main study, provided the names of two professors and two classmates to be embedded in the preambles, as in the main experiment. They evaluated the imposition, power, and distance in each situation using a 9-point Likert scale. For imposition, the endpoints of the scale were pedir poco ‘undemanding’ or pedir mucho ‘over-demanding’ (Figure 2). For power, the scale ranged from iguales ‘equals’ to desiguales ‘unequals.’ They rated distance in terms of closeness, from relación cercana ‘close relationship’ to relación distante ‘distant relationship’.

The perception of each variable was analyzed separately, using a hierarchical regression to control for scoring variability among raters (modeled as varying intercepts), other factors (e.g., different power levels when scoring distance), and the interactions. The norming analyses showed a difference of 5.07 (p < 0.001) between low- and high-imposition situations; a difference of 4.35 (p < 0.001) between low- and high-power situations; and a difference of 5.44 (p < 0.001) between low- and high-distance situations. The results suggest that the Spanish native speakers perceived the situations to describe the intended distinctions of power, distance, and imposition.

3.3. Data Coding and Analyses

All responses were coded according to the verb form in the head act of the request (i.e., the part that communicates the speech act). Considering the prior literature on verbs and politeness (Chodorowska-Pilch 1998; Félix-Brasdefer 2005; Koike 1989) and the current data, the following verb forms were coded: subjunctive; modal conditional with the verb poder ‘to be able to’; conditional, past, future, modal present with the verb poder ‘to be able to’; present; and imperative. Of 1776 responses, a total of 239 responses were eliminated. In 124 responses, no request was performed, and no verb was used in 4 situations. Additionally, the 111 responses from situation 10 (Appendix A) were eliminated because the situation did not prompt participants to make a request in the present moment as the others did. The gender of each participant and the contextual variables present in each situation were also coded for all responses.

To respond to the research questions, a mixed-methods approach was used to examine the data quantitatively and qualitatively. In the first part of the analysis, two separate conditional inference trees were fitted to female and male participants’ data with verb form as the dependent variable (eight levels) and the predictor variables of power, distance, and imposition (two levels each) as independent variables. Conditional-inference-tree analyses determined the level of significance and the hierarchical order of each of the predictor variables on verb form by gender. Conditional inference trees are a type of random forest analysis that implements tree-structure regression models into the framework of conditional inference procedures (Hothorn et al. 2006). Within each conditional inference tree, statistical tests are performed to determine whether each split of the tree is significant or not. In addition to utilizing and providing tests of significance, the output, based on machine learning algorithms, provides a visualization of how predictors operate. The model was constructed by using the party package and the function ctree() (Hothorn et al. 2006) in R (R Development Core Team 2019). Similar statistical approaches have been used in previous sociolinguistics and pragmatics research (e.g., Sainzmaza-Lecanda and Schwenter 2017; Tagliamonte and Baayen 2012; Tamaoka et al. 2010; Rosemeyer and Schwenter 2017).

Following the conditional-inference-tree analyses, the distribution of verb forms by women and men was analyzed using a chi-square test for independence. Then, individuals’ verb-form-selection tendencies were visually examined using a ridgeline plot. The unique stances adopted by each participant were also analyzed, relying on pragmatic interpretations of verb forms as communicating more or less directness with respect to the request. The benefit of adopting the notion of stance, even when the underlying concept of directness has been addressed in the prior pragmatics literature, is that it draws attention to the creation of social meaning in given interactions rather than relying on a set system of pragmatic meaning applied uniformly by all speakers. Excerpts are unaltered to reflect the original wording of the participants.

4. Results

4.1. Effects of Power, Distance, and Imposition on Verb Selection by Gender

Providing the best fit hierarchical structure, the results of two conditional inference trees show how the predictor variables (i.e., power, distance, imposition) affect verb selection in Peninsular Spanish requests for women and men in Madrid, Spain. The inner nodes represent the effect of the different predictor levels and demonstrate their significance, and the terminal nodes at the bottom of the output show the distribution of the verb forms used.

The conditional inference tree for the women’s data indicated that all predictor variables examined had a significant impact on verb selection (see p-values in Figure 3). Power was the highest node (p < 0.001) (Node [1]). When a power differential was present (+power), distance was found to be the next-highest-ranked predictor (p < 0.001) (Node [2]), and imposition only had an effect when requests involved a person with whom the participant was familiar (−distance) (p < 0.001) (Node [4]). On the other side of the tree, for the situations with equal power (−power), the predictor of imposition was ranked as the next-highest-level predictor (p < 0.001) (Node [7]), followed by distance (p = 0.004 and p < 0.001) (Nodes [8] and [11]).

The conditional inference tree for the men’s data also showed that all predictor variables had a significant impact on verb selection (see p-values in Figure 4). In Figure 4, power (p < 0.001) was found to be the main predictor of verb selection (Node [1]), as in the model of the women’s data. When there was a power differential (+power), imposition (p < 0.001) was found to be the next-most-important predictor (Node [2]), with distance being a significant predictor only when the requests involved low imposition (−imposition) (p < 0.001) (Node [4]). On the other side of the tree (−power), imposition (p < 0.001) was the next-highest-level predictor (Node [7]), followed by distance (p = 0.006 and p < 0.001) (Nodes [8] and [11]).

Examining the inner nodes in Figure 3 (women’s data) and Figure 4 (men’s data), the rather high placement of imposition across both models is notable, considering that prior research has focused more on power and distance than imposition. Furthermore, the comparison across the women’s and men’s data highlights the inverted hierarchical order of distance and imposition in contexts of +power, with women relying on distance and men relying on imposition as a higher-level predictor. The difference in the hierarchical structure provides evidence that the way in which women and men in Madrid rely on the contextual variables to make linguistic choices is different, at least when making requests, as represented by the current data.

4.2. Distribution of Verb Forms

To understand the overall distribution of verb forms by gender, the counts of verb forms used by women and men were calculated (

Table 2. Count and percentage of verb forms by women and men.

Verb Form	Women		Men
	n	%	n	%
Subjunctive	92	9.6%	38	6.6%
Modal Conditional	279	29.0%	156	27.2%
Conditional	110	11.4%	83	14.5%
Past/Future	24	2.5%	18	3.1%
Modal Present	139	14.5%	81	14.1%
Present	232	24.1%	126	22.0%
Imperative	86	8.9%	72	12.5%
Total	962	100%	574	100%

) and a chi-square test of independence was performed.9 The results indicated independence between gender and verb form, X²(6, 1536) = 12.55, p = 0.05. While both groups used a wide range of verbal forms and relied most on modal conditional and present tense verb forms in requests, they exhibited slightly different verb form distributions. Women used subjunctive, modal conditional, and present slightly more often than men, while men used the conditional and imperative forms slightly more than women. Considering the verbal continuum, women used slightly more verb forms on the higher end of the continuum (e.g., less direct), and men used slightly more forms on the lower end (i.e., more direct).

An examination of the verb forms that were most distinct by group in the terminal nodes of the conditional inference trees revealed that women employed the subjunctive more frequently than men across situations. Men used imperative forms more than women, particularly in −power, −distance, and −imposition situations (66.7% for men and 47.3% for women). In this specific situation, women were more likely to use a present tense verb than men—another indicator of the women’s slight tendency to select forms towards the higher end of the continuum compared to men.

4.3. Individual-Level Analysis

To examine individuals’ verb choices and related stances, each participant’s set of requests was analyzed with respect to their position on the verbal continuum. Quantitatively, a mean score was calculated for each participant by assigning a 1–7 scale to the verbs on the verbal continuum (1—subjunctive, 2—modal conditional, 3—conditional, 4—imperfect/future, 5—modal present, 6—present, 7—imperative).10

The ridgeline plot in Figure 5 shows the distribution of the means by gender. While the group means for women (M = 3.93, SD = 0.67) and men (M = 4.06, SD = 0.75) are not notably different, the plot demonstrates various trends related to individuals’ verb-selection tendencies. First, there is a similar distribution for women and men, with most data points appearing near the center of the continuum. This finding and the distribution data reported in Table 2 indicate that most participants rely on some verb forms on the high and low ends of the continuum, given that the midpoint past and future forms are seldom used. This trend is supported by the conditional-inference-tree analyses, which demonstrate that the use of verb forms on the high and low ends of the continuum aligns with specific contexts. Second, the ridgeline plot also highlights the wide range of individuals’ means represented in the data, which ranged from 2.46 to 5.78 for women and 2.36 to 6.07 for men, indicating individual tendencies in verb selection for certain women and men that deviate from each group’s norm. Third, whereas the women’s data show a more normal distribution, the men’s distribution tends towards verb forms on the low end of the spectrum, using verbs that are more direct or assertive. This finding is seen in the small increase in data points between 6 and 7 on the scale and in the larger set of data points between 4 and 5. There also seems to be a subgroup of men that prefer to use verb forms that are higher on the continuum than the mean, indicated by the increase in data points between 3 and 3.5. Overall, these data confirm that there is a slight tendency for men to use more-direct verb forms than women, but it also highlights the great range in variability among individuals.

To explore the variability in verb form use, data from individual participants were selected for analysis and presented in Section 4.3.1, Section 4.3.2 and Section 4.3.3. These data show that some men and women fall outside of the general and gendered norms in terms of their use of verb forms.

4.3.1. Midpoint on the Verbal Continuum: Stance Varies by Situation

The participants with individual means that aligned with the group mean tended to use verb forms across the continuum. With the infrequent use of past and future verb forms overall (Table 2), a midpoint average was achieved by using a mix of verb forms from both ends of the continuum. Demonstrating this trend, Sofía (Participant 104) had a mean of 4.0, aligning with the women’s mean. In the 15 situations, she used three modal conditional verbs, four conditional verbs, one past verb, four modal present verbs, and three present tense verbs. Jorge (Participant 77) had a mean of 4.2, approximately aligning with the mean for the men’s data. In his requests, he used all verb options except the future: one subjunctive form, two modal conditional forms, three conditional verbs, two past forms, two modal present forms, four present tense verbs, and one imperative.

For both participants, there was an alignment between verb form and situation. For example, they used verb forms on the high end of the continuum in the +power, +distance, and +imposition situations and verb forms on the low end of the continuum in the −power, −distance, and −imposition situations. In +power, +distance, and +imposition situations, Sofía used modal conditional verbs (Excerpts 9 and 10), whereas in −power, −distance, and −imposition situations, she used a modal present and a present tense verb (Excerpts 11 and 12). Similarly, Jorge relied on verb forms towards the high end of the continuum in the +power, +distance, and +imposition situations, where he used a subjunctive and a past form (Excerpts 13 and 14), while in the −power, −distance, and −imposition situations, he used a present tense and imperative verb to perform the requests (Excerpts 15 and 16). These participants and others whose data represent the average for their group used a wide range of verb forms, more frequently employing forms on the high end of the verbal continuum in +power, +distance, and +imposition situations and forms on the low end in the −power, −distance, and −imposition situations.11

(9)

Sofía. Modal Conditional (+power, +distance, +imposition).

Me gustaría optar a una beca, crees que podrías hacerme una carta de recomendación o sabes de alguien que pueda hacermela?

I would like to apply for a scholarship, do you think you could give me a letter of recommendation or do you know of someone who could do it for me?

(10)

Sofía. Modal Conditional (+power, +distance, +imposition).

Cómo podría conseguir el manual?

How could I get the manual?

(11)

Sofía. Modal Present (−power, −distance, −imposition).

Te puedo coger un folio?

Can I take a sheet of paper from you?

(12)

Sofía. Present (−power, −distance, −imposition).

Lo tiras plis?

Throw it away please?

(13)

Jorge. Subjunctive (+power, +distance, +imposition).

Buenas tardes profesor. El motivo de venir es pedirle un favor de motivo académico. Estoy interesado en solicitar una beca… Como considero que usted es uno de los especialistas de mayor renombre en nuestro país en ese campo, quisiera pedirle por favor que me redactara una carta de recomendación para poder solicitar dicha beca…

Good afternoon professor. The reason for coming is to ask you for an academic favor. I am interested in applying for a scholarship… Considering that you are one of the most renowned specialists in our country in that field, I would like to ask you to please write me a letter of recommendation to be able to apply for the aforementioned scholarship…

(14)

Jorge. Past (+power, +distance, +imposition).

Disculpe profesor. Quería pedirle los datos de referencia de su manual para poder estudiarlo para su asignatura. Muchas gracias.

Excuse me professor. I wanted to ask you for the reference of your manual so that I can study it for your class. Thanks very much.

(15)

Jorge. Present (−power, −distance, −imposition).

Miguel, ¿me dejas un par de folios?

Miguel, will you give me a few sheets of paper?

(16)

Jorge. Imperative (−power, −distance, −imposition).

Tío, Miguel, tírame esto a la papelera por favor.

Dude, Miguel, throw this in the waste basket for me please.

The verb forms that have been described as more or less direct in pragmatic terms index how individual interlocutors are positioned with respect to each other and the request (i.e., stance (Du Bois 2007)). They also communicate pragmatic meaning about the deictic point of occurrence, placing the action in a more- or less-distanced or irrealis moment. This pragmatic meaning has social implications, as noted in prior research on politeness (Chodorowska-Pilch 1998, 2004; Koike 1989). Selecting a more-direct verb form contributes to the stance of a ‘direct requester’. ’Direct requesters’ position themselves socially towards the hearer, as being in a relationship that requires minimal worry about impeding on the hearer, there is a greater assumption that the hearer is willing and able to perform the requested action (Searle 1976). Additionally, they position the speaker and hearer around a request that is close in terms of time or reality. The stance of a ‘direct requester’ can be seen in Excerpts (12), (15), and (16). On the other hand, the stance of an ‘indirect requester’ highlights caution around the degree to which the speaker assumes that the hearer is able or willing to perform the request, thus expanding the imagined, socially positioned space between the speaker and hearer. Indirect verb forms also present the request itself as more distanced or removed from the reality of the speaker and hearer.

In addition to viewing directness and indirectness as a stance-taking mechanism, it is important to note that the stances displayed by participants at the midpoint of the range are broadly sensitive to the contextual variables. This set of participants shows that they sometimes behave as ‘direct requesters’, especially when they are in −power situations, and they behave as ‘indirect requesters’ at other times, especially in +power situations. The sensitivity to contextual variables aligns with hypotheses based on pragmatic theories and the prior literature about verb forms and contextual variables.

4.3.2. High End of the Verbal Continuum: Indirect Requesters

Some individuals used verb forms on one end or the other of the verbal continuum. Elena (Participant 49) had an average of 2.71, aligning her verb forms across situations with the high end of the continuum, which expresses more-indirect requests. She used subjunctive forms two times, modal conditional forms in six requests, conditional verbs four times, and present tense verbs in two situations. Similarly, Rafael (Participant 25) had an average of 2.36. In his requests, he used modal conditional verbs in 12 situations, and he used a conditional verb and a present tense verb one time each. These participants used verbs to make requests in a way that is different from those who relied on the full spectrum of forms; they more consistently employed forms that theoretically have been proposed to be less direct and communicated with less force. Excerpts (17)–(19) show Rafael’s application of a modal conditional verb in a range of different situations with different combinations of contextual variables. A connection between situation type and verb form was not clearly observable. Thus, it seems that some individuals prefer verb forms in requests that more consistently present them as ‘indirect requesters,’ conveying tentativeness about the assumption that the hearer is willing and able to perform the request and by distancing it from the interlocutors via deictic displacement.

(17)

Rafael. Modal Conditional (+power, +distance, +imposition).

Gómez, necesito solicitar la beca para el año que viene. Podrías hacerme una carta de recomendación, por favor? Gracias

Gómez, I need to apply for the scholarship for next year. Could you give me a recommendation letter, please? Thanks.

(18)

Rafael. Modal Conditional (+power, −distance, −imposition).

Elena, ¿Recuerdas los artículos que nombraste en clase? ¿Me podrías decir cuáles eran, por favor?

Elena, do you remember the articles that you named in class? Could you tell me which ones they were, please?

(19)

Rafael. Modal Conditional (−power, −distance, −imposition).

Pablo, me podrías dejar unos folios, por favor?

Pablo, could you give me some sheets of paper, please?

4.3.3. Low End of the Verbal Continuum: Direct Requesters

The data also indicated that some people prefer to use verb forms in requests that align with the low end of the continuum—verb forms that are more direct, such as imperative and present tense verbs. María Carmen (Participant 85) had an average of 5.47. She used one modal conditional form, three conditional forms, six present tense verbs, and five imperative verbs in her requests. Alberto (Participant 97) had an average of 6.07 on the verbal continuum, and he used one modal conditional, two modal present, four present, and seven imperative verb forms.

María Carmen used imperative forms in various situations, from those with a low-imposition request to her professor, with whom she was not very familiar (Excerpt 20), to those that involved a low-imposition request to her close peer (Excerpt 21), but her use of the imperative was not limited to situations of low imposition. She relied on present and imperative requests in many more situations than most participants, and she showed minimal sensitivity to the contextual variables. Similarly, Alberto’s data did not show any observable distinction across situations. Both María Carmen and Alberto relied quite heavily on the most-direct verb forms, using the present or imperative in 11 of the 15 situations, indicating a quite consistent ‘direct requester’ stance. This stance was communicated by maintaining the request within the referential time period of the present/realis moment and by not indicating that they questioned the assumption that the hearer is willing and able to carry out the requested action.

(20)

María Carmen. Imperative (+power, +distance, −imposition).

Qué se ha escrito sobre este tema? Recomiéndame algo bueno.

What’s been written about this topic? Recommend something good for me.

(21)

María Carmen. Imperative (−power, −distance, −imposition).

Tíralo, por favor.

Throw it out, please.

The qualitative samples of data in Section 4.3.1, Section 4.3.2 and Section 4.3.3 confirmed that not all participants behaved in the same way. While most of the participants were similar to Sofía and Jorge, who used verb forms on the high and low ends of the verbal continuum in accordance with the contextual variables, as predicted by prior research, some participants preferred a less-direct stance, like Elena and Rafael, and some preferred a more-direct stance, like María Carmen and Alberto. Taken collectively, these data suggest that many participants adopt direct and indirect stances in response to contextual variables, as expected (e.g., more direct with −power, less direct with +power), but others take on direct or indirect stances more consistently and with less dependence on contextual variables. The participants who are more responsive to contextual variables take different stances depending on the context in systematic ways, while those who are more consistent in their stance-taking maintain more uniformity in how they position themselves when making requests. These findings give rise to new questions about why some individuals are more affected by contextual variables than others and how the outliers may shape the future norms of a given population.

5. Summary of Findings and Discussion

This investigation examined how contextual factors of power, distance, and imposition collectively condition verb forms used by women and men in Madrid, Spain, and how verb selection varies by gender and by individual. In this section, a response to each research question is provided, followed by a discussion of the main contributions.

5.1. Effect of Contextual Variables

In response to the first research question about whether the effect and hierarchical order of the contextual factors of power, distance, and imposition differed for women and men when predicting verb forms in requests, the results indicated that there was a difference in the effect and hierarchical order across groups. The women’s data showed that distance was a more important predictor than imposition in situations of asymmetrical power (+power). For the men’s verb selection in the same asymmetrical power situations, imposition was a more important predictor than distance.

The difference in the hierarchical structure of predictor variables demonstrates the variability in how certain social groups rely on contextual factors to make linguistic choices. Whereas Tamaoka et al. (2010) showed that particular cultural groups differentially value power and distance when interpreting politeness, the current data show that, beyond the cultural level, individual social groups can interpret context differently. The women and men who participated in the current experiment were all exposed to the same situations and yet the results point to different ways of relying on the contextual variables to make linguistic choices.

The conditional inference trees for women and men provide evidence of different frames (Terkourafi 2005, 2015). Within these unique frames, women’s verb-form selections in situations with an asymmetrical power relationship (+power) are more tightly connected to whether the hearer is a more- or less-distant person; men’s verb forms in those same situations are more tightly connected to the degree of imposition of the request. Thus, even when two individuals select the same verb form in the same situation involving a power differential (+power), the results demonstrated that their frame-based meanings may be distinct. In other words, when the predictive hierarchies are different, the verb forms utilized by the women and men whose data contributed to the overarching results theoretically index diverse positionings around notions of distance or imposition. These associations between contextual variables and linguistic resources, which are based on the interlocutors’ predictive abilities about frames (Terkourafi 2015), provide nuanced shifts in the social meaning of the linguistic resources used. Whereas variationist approaches have been common in linguistics research for decades, mostly focusing on the variation in linguistic resources (e.g., subjunctive vs. indicative; [s] vs. [h]), the current findings draw attention to the possibility that the perception of the context is also a variable factor that impacts linguistic production and meaning.

The results of the conditional-inference-tree analyses also shed light on a more basic question about the importance of power, distance, and imposition for understanding language variation. Of the three, power has perhaps been the most studied in pragmatics research (e.g., Félix-Brasdefer 2005, 2009; Hernández-Flores 2004), and theoretical discussions have also pointed to power being a primary influence on language and society (e.g., Eckert 1989; Uchida 1992). While distance has also been considered frequently, less systematic attention has been paid to imposition (e.g., Ballesteros 2001; Blum-Kulka et al. 1989; Félix-Brasdefer 2005, 2009; Le Pair 1996), under the assumption that it is a less important contextual variable. The results of the conditional-inference-tree analysis confirmed the importance of power, and they demonstrated that imposition was often a more influential variable than distance when predicting verb forms in requests. In an applied sense, this finding suggests that the Spaniards who participated are more sensitive to imposition than may have been expected, and future research should consider imposition when examining the effect of contextual variables.

5.2. Verb Form and Gender

The second research question addressed the distribution of verb forms used by women and men. Chi-square results indicated a significant difference in the distribution of verb forms across women’s and men’s requests. The results provided some support for the notion that women may prefer to use softened linguistic forms more than men (e.g., Leaper and Robnett 2011). In the current data, women used subjunctive and modal conditional forms slightly more than men, and men used imperative verbs—the most-direct verb form examined—more than women and in more situations. These observations did not align with Lorenzo-Dus and Bou-Franch’s (2003) finding that Spanish women tended to use more-direct requests than men in situations with asymmetrical power and distance; rather, the current results indicate subtle differences in the selection of verb forms at the extremes of the directness continuum associated with politeness (i.e., subjunctive, modal conditional, imperative).

5.3. Individual Stances with Verbs

The final research question addressed the variability in individuals’ verb forms and the stances related to directness that they portray. Most men and women used varied verb forms that reflected more- and less-direct stances in different situations (e.g., more direct with −power, less direct with +power). Most participants’ range of verb forms was fairly equally distributed across the verbal continuum, although some participants maintained a more- or less-direct stance overall. These participants acted more as ‘direct requesters’ or ‘indirect requesters’ independent of the situation. The analysis at the individual level aligned with the understanding that macro-social trends are representative of large groups of individuals who share common linguistic practices (e.g., Bell 2016), and it also confirmed the agency of individuals. As clearly indicated in third-wave sociolinguistic research (Eckert 2012), the analysis of individuals’ language use is fundamental for understanding the social moves that are made by all speakers, not just those that fall within the norm. In other words, the ‘direct requesters’, ‘indirect requesters’, and the rest of the participant population who relied on contextual variables to a greater degree to select verb forms are all positioned vis-à-vis each other; the social meaning of these performances is best interpreted by considering the relationship to the other performances. Furthermore, the social moves performed via language, including those made by the outliers in these data, have the potential to affect the request norms and the continued (re)construction of gender in Peninsular Spanish.

While the individual agency was apparent, the current data were not sufficient to explain what ideologies, stances, identities, or frames explain the individual-level variation. For example, it is possible that participants who represent the outliers have different underlying hierarchies when processing contextual variables or different responses to them. It is also possible that these explanations, related to the perception of context, may be intertwined with speakers’ ideologies and identities.

Reflecting on the results as a whole, one may question the degree to which interlocutors are aware of their linguistic choices that rely on interrelated contextual variables or the nuanced meanings that emerge from the coupling of a linguistic form with a specific interpretation of a context. Eckert (2019) proposes that ‘most of what we do …, we do unconsciously’ (p. 758). She suggests that speakers may be more aware of the persona that they are trying to communicate rather than the individual features of a style. Following this line of argumentation, it seems unlikely that speakers or hearers would be conscious of the meaning contributed by the hierarchical structure of predictive contextual variables, but this does not mean that it is not unconsciously calculated. In other words, they may consider themselves to be ‘direct communicators’, ‘indirect communicators’, or even communicators who do not like to impose or value close relationships, but they may not be aware of how they use linguistic resources or how they rely on contextual variables when constructing these positionalities.

6. Conclusions

This investigation examined the intertwined impact of power, distance, and imposition on verb selection in Spanish for women and men in Madrid, Spain, and explored individual strategies in the formulation of requests. The results of this study showed differences across the women’s and men’s data in the hierarchical structure of the three contextual variables as predictors of verb form, which highlights contextual variability as a factor that should be considered in the study of the social meaning of language. Other findings demonstrated differences in verb use not only between Spanish women and men but also among individuals. Taken as a whole, the results of this investigation are a snapshot of the participants’ complex calculations involving language, context, and interlocutors, and they support the notion that frames, politeness, and gender are domain-specific, dynamic constructs in communities (Mills 2003; Terkourafi 2015).

The experimental paradigm and mixed-methods approach in this study offered the benefits of isolating specific contextual variables, collecting a large amount of data in controlled situations, and providing new insight into the complex relationship among linguistic variables, contextual variables, and social groups and individuals. Despite these strengths, the experimental data were also limited in that they do not represent naturally occurring speech. Another limitation was the scope of the investigation, which was focused on verb forms. Future research should consider other types of data to corroborate and further the current findings. For example, while prior research has shown that the verb is a fundamental indicator of illocutionary force and politeness (Chodorowska-Pilch 1998, 2004; Félix-Brasdefer 2005), future research may seek to confirm the hierarchical structure of contextual variables through the analysis of other linguistic forms, as they may be predicted by the same or a distinct organization of contextual factors. This line of research will continue to demystify the understanding of language and context. Additionally, while the current project determined that some individuals adopt more-direct or -indirect strategies regardless of context, future research should explore the social explanations for these results. Analyses of metalinguistic data or ethnographic data related to individual participants could provide necessary insight to explain the local and social meanings of the linguistic resources in use.