Relationships as regulators of discourse interaction in Spanish

In this paper we argue that in analyses of discourse interaction the relationship between participants must be considered as a parameter that conditions their linguistic behavior. A relationship is the result of the experiences shared by partners which create a “cultural code” between them. Each relationship brings a set of circumstances into play that regulates how members behave and how they interpret each other’s words and actions. Our proposal stems from observing how Spanish speakers who hold different types of relationships handle opposing points of view in conversation. Our results show that the expression and interpretation of dissent is linked to the interlocutors’ need to maintain the relationship in good terms. In order to protect relationships, the basic principle guiding interaction is to reach consensus. Additionally, the type and strength of the relationship condition the impact of expressions. Strong disagreement is allowed in strong relationships, as the risk for its affecting the relationship is low. Conversely, weak disagreement is preferred in weak relationships, since opposing views may result in higher impact.


Introduction
Recent research in Interpersonal Communication and discourse analysis (Spencer-Oatey 2002Mandelbaum 2003;Pomerantz and Mandelbaum 2005;Enfield 2006Enfield , 2009Locher and Graham 2010;Arundale 2010) has emphasized the importance of the relationships between interlocutors as a regulatory tool of communicative exchange in interaction.
The concept of relationship is not new. Sociolinguists have incorporated this variable since its inception (See Milroy 1987, among others). Also, pragmatic _____________ 1 Universidad Pública de Navarra. Correo electrónico: magdalena.romera@unavarra.es studies have stressed the importance of social rules in the use of linguistic strategies by interlocutors (Fraser and Nolen 1981;Gu 1990;Watts 1992;Chen 1993). Conversation Analysis has done so even more extensively (Mandelbaum 2003;Pomerantz and Mandelbaum 2005). However, as Mandelbaum (2003) or Arundale (2010) argue, most studies have incorporated this variable from a weak point of view, i.e. understanding the "social relationship" as the sum of the specific characteristics of individuals in a communicative act, or something that individuals have, instead of as an integrative concept of the realities of its members, which results in a new set of circumstances dynamically constructed in interaction. This latter view is what Arundale (2010) called the "strong version" of the relationship.
In this paper, we rely on the latter version of relationships. The experiences shared by the participants create a new "cultural code" between them (Burleson et al. 2000), a social and linguistic code which becomes exclusive for the members of the dyad. These codes govern communication between interlocutors, beyond the norms of the community of practice and socio-cultural norms.
In the characterization of relationships, typology is important. As Pomerantz and Mandelbaum (2005) suggest, "participants in interaction use their knowledge of the activities, motives, rights, responsibilities, and competencies that they regard as appropriate for incumbents of a relationship category, for example, as appropriate for friends, mothers, or children." (149). However, characterizing relationships only on the basis of categories does not account for how in similar situations, identical linguistic expressions do not produce the same communicative effects.
Consequently, types of relationships are one important part of this concept, but participants in each relationship also behave according to their own "exchange rules", i.e. specific social and linguistic parameters that characterize a relationship. These rules determine the interlocutors' limits in a particular interaction above other linguistic and social factors. This also explains why, in the absence of a relationship, interlocutors lack their own codes and are guided by higher level rules of behavior, i.e. cultural norms for categories (mothers, siblings, friends, etc.) which are acceptable in each society, according to the goals of the participants (Bousfield 2008), and the rules of exchange of the community of practice (Wenger 1998) the speakers belong to.
As Enfield (2006) suggests, our choice of linguistic expressions is not only made in terms of informativeness, but also according to what it represents in the relationship linking us to our interlocutors. Information must be conveyed as well as possible so as to transmit the intended meaning, but we, as speakers, must also make sure that we convey it appropriately to maintain our relationships in good terms. Both imperatives, the informational and the affiliational, are carefully respected in discourse. Mandelbaum (2003) shows how a speaker may choose to use an identical replica, or a repair to deactivate an intervention that compromises the relationship. Enfield (2006) also shows how in the use of referential expressions, informativeness can be reduced to benefit the recognition and maintenance of the relationship. Other parts of speech can also be analyzed in the same terms. In this article, we focus on observing how the expression of dissenting opinions is conditioned not only by what we need to convey, but also by the relationship held between the participants. We especially focus on uncovering how maintaining the relationship in good terms is one of the primary objectives in discourse, just as important as being informative.
We are aware of the methodological complexities of this study. Characterizing relationships and establishing a correlation with the linguistic expressions used is a hard task. The fact that each relationship establishes its own codes impedes a systematic approach and makes it necessary to rely on the analyst's interpretation of the speaker's behavior, which of course entails problems of validity (Cf. Mandelbaum 2003). This is why in this study we use data from semi-structured conversations, where all verbal and nonverbal factors were carefully coded following discourse analysis conventions (Mandelbaum 2003;Pomerantz & Mandelbaum 2005;Bousfield 2008;Arundale 2010;Culpeper 2011;Enfield 2006Enfield , 2009Stivers 2008, among others).

Relationships
Humans are social beings and as such they need to interact with others to survive. So important are group membership, social acceptance and recognition that individuals typically strive to maintain active and healthy links with members of their community. Relationships are therefore essential for survival, and human social behavior will aim to preserve them (Agha 2006).
As we said, we agree with the strong version of relationships (Arundale 2010), in which a relationship is seen as a dynamic and constantly changing entity which develops through interaction. Relationships are the result of the common experiences shared by their members that build a specific set of circumstances for them.
Their continuous interaction creates a "cultural code" for them. This is what Burleson, Metts and Kirch (2000), following Goldsmith and Baxter (1996), have called "relational culture": "shared meaning systems […]; characteristic interaction routines and rituals; norms and rules that organize, sequence, and control behavior; and role structures that organize situated identities […]." (252).
Shared experiences generate specific cultural, social and linguistic codes among speakers. Members of a relationship create common shared meanings for elements such as people, objects, events, etc. They also create specific terms to designate important aspects for their members. In each relationship there are also standards of behavior agreed upon by the actants or accepted by society, which prescribe the limits of the relationship. In this paper, we refer to all these parameters as the relationship's "exchange rules". They govern the interaction in that relationship and are inserted into the socio-cultural rules and into those of the community of practice. These rules determine the greater or lesser impact of linguistic expressions within the relationship. The boundaries of the socio-cultural norms and norms of the Community of Practice are fairly rigid and difficult to change. Exchange rules, created through successive exchanges however, are much more flexible and customizable; their limits can transcend social norms, but usually do not exceed the limits of the general plane or macro social level.
A relationship grows from the historical elements that condition the link between the individuals, i.e. the origin of the relationship, its length, and the degree of intimacy between members. Some relationships are created; others come established beforehand. The interaction defined by an organization such as family relationships, neighborhood or professional ties (Cf. Degenne 2009), condition the subjects to behave in a pre-established manner. Not all relationships develop in the same direction or to the same extent; some are bound to be unstable and temporary, while others grow in strength with time. The length of the relationship and the degree of intimacy both restrict the way the relation moves. The degree of intimacy results from the strength of the shared history and common past interactions (Ferrand 2006;Bourdon 2009). Shared experiences can strengthen a relationship and help create a reciprocal commitment (Finch and Mason 1993;Degenne and Forsé 2004;Grossetti 2005;White 2009), by which both partners are expected to bring reciprocity (Gouldner 1960;Degenne 2009) into the relationship. This affects the social behavior, as well as the linguistic behavior that the interactants are expected to follow. The greater the strength of the relationship, the higher the commitment to each other. Factors relating to the social hierarchy of the speakers also condition the exchange rules in the relationship. The status and power among interlocutors are settled according to their social position within their community and the dependence relationships established between them. They both determine the symmetry of the relationship and the behavior of the speakers towards each other.
Interactions are also governed by situational variables: the purpose of the interaction, the speakers' needs, and the characteristics of the participants (age, sex and level of education) all come into play. Finally, besides these, the individual conditions of the actants must also be taken into account. Individuality is conditioned by personality and temperament, and it reflects a personal way of acting and interpreting things. Individual characteristics shape the way interactants relate, and set the exchange rules for that relationship. As Goldsmith and Baxter (1966) or Enfield (2009) argue, dialectic forces, i.e. negative and positive forces, participate in every relationship. Not only are trust and common identity promoted, but competition, assertiveness and distinction are also at stake. The individual characteristics of each subject play an important role here. Due to the balance of these forces, conflicts occur and are particularly threatening when they affect the relationship's future.
Given all these variables, as noted by Degenne (2009), when interaction occurs for the first time, actants begin setting up their own relational rules. Initially, they will respect the social-cultural norms set in their society and their community of practice (Marra 2012). As their mutual understanding and their shared common experiences increase, their exchange rules will get more specific and become the ones they follow, i.e. consecutive encounters will determine the course of the relationship and will classify certain behaviors as admissible and/or expected from its members.

Relationships and disagreement
As we said, the maintenance of relationships becomes an intrinsic and constant existential principle for human beings, who always act towards preserving them (Agha 2006). Conflicting opinions in everyday interaction, however, seem a priori to go against this principle, since they can be considered a divergent strategy leading to detachment from others. The expression of different points of view generates situations that may threaten a relationship, since it creates more distance than approximation. Our hypothesis, however, defends that their impact clearly depends on the relationship held by participants. Our work shows that the exchange rules established by interactants, or alternatively the absence of them, determine both the acceptability and effect of expressions in conflict.
Disagreement in interaction has been studied in the framework of impoliteness research (Leech 1983;Brown and Levinson 1987;Kasper 1990;Waldron and Applegate 1994;Herrero Moreno 2002;Brenes Peña 2011). Early politeness studies considered disagreement to have an inherently negative and impolite impact (See Sifianou 2012). Avoidance of conflict was one of the main regulatory maxims of politeness at the time (Leech 1983;Kasper 1990), so dissent was relegated to the group of potentially conflicting and threatening speech acts. More recent research, however, has surpassed this conception of the inherently impolite nature of dissent. Disagreement is no longer considered an isolated speech act, but is comprehensively analyzed within the overall production of discourse (Eelen 2001;Watts 2003;Spencer-Oatey 2005). To measure the impact of an act of disagreement, new variables from the context and the co-text are now taken into account, such as the overall management of discourse, intercultural and intracultural variation, the individuals' personal traits, interactions previously held by the participants or the multifunctional and polysemous nature of linguistic expressions (Kakavá 1993;Hayashi 1996;Locher and Watts 2005;Limberg 2009;Sifianou 2012).
We endorse this conceptual consideration of the expression of disagreement. However, accepting that its perception and level of impact are conditioned by the social conventions of the community, we argue that the relationship is of particular relevance when setting the limits of interpretation, i.e. on the acceptance and impact of any expression in interaction. This hypothesis is already outlined in classic studies on disagreement as well. Since Schiffrin's (1984) work on disagreement among friends, it has been accepted that disagreement may create affiliation in intimate relationships. We suggest that, as Georgakopoulou (1998; successfully argues, disagreement is allowed because it does not pose a threat to the relation. In this way, the occurrence of disagreements does not seem to pose a threat to the participants' relation. In fact, the group's ethnographic study suggests that participants base their friendship on openness and the freedom to debate various issues. As they frequently state, 'there are no hard feelings involved in speaking our minds and telling each other off, when we feel we should; what are friends for anyway?'. Similarly, their conversations foreground their 'best friends' identity and its concomitant elements of intimacy and camaraderie. Interactional history and shared assumptions are also part and parcel of the participants' closeness (Georgakopoulou 2001:1897) Disagreement occurs primarily between two participants (Clayman, 2002) and due to its implications, speakers assess the disagreement according to their own interpretation of the other speaker's communicative intentions and their relationship. If dissent has caused some undesired impact, then it is up to the receiver to react accordingly. Classifications in the literature on responses to conflict (van de Vliert and Euwema, 1994;Hojjat, 2000;Canary 2003) suggest two ubiquitous dimensions over which answers range: 'directness' and 'cooperation'.
The 'directness' axis indicates how much the speaker gets involved in the response, whereas 'cooperation' signals whether the individual's attitude is cooperative or competitive. In accordance with these distinctions, in this paper, we follow Bourdon's (2006) taxonomy, which suggests three possible reactions.
1. Avoidance: Ignore the discrepancy to avoid conflict. 2. Negotiation: Negotiate or approach to gain compliance. 2.1. Converge: attempt to approach the opposite position. 2.2. Make the other Converge: try to attract the other towards our position. 3. Separation: separate oneself from the opposite position, using noncooperative interaction (Canary 2003). But what will make the listener lean toward one type of resolution over another? Besides other factors (Vid. Canary 2003), our hypothesis rests on the value of the basics principles of Communal Sharing (Fiske el al. 2005) and Affiliation (Bravo 1996;Enfield 2006;. These two principles state that the most natural option in human interaction is to seek common ground and reach consensus, since this tends to build and maintain relationships. Relationships develop on consensus and cooperation. When partners interact for the first time, interventions are especially cooperative and try to seek common ground. Dissent in initial meetings is not well tolerated, for it is interpreted as a sign of underestimation and imposition on the other's opinions. We are aware that this vision emanates from the classic Gricean thinking that communication is always governed by cooperation (Garcés-Conejos, 2013). However, as numerous studies on impoliteness (Bousfield, 2008;Culpeper, 2011; among others) have shown, conflict and impolite behavior occur pervasively in daily encounters as well. Bearing this in mind, we still consider that even in confrontational argumentative contexts such as political or academic debates, where interlocutors are expected to express opposing opinions overtly, the goal is to make one of the positions prevail. Each interlocutor tries to win over his/her opponent by undermining their arguments, and both try to achieve consensus by making the other converge. Burleson et al. (2000: 247) propose three necessary tasks that speakers should manage for a successful interaction. These are: the instrumental tasks, or those defined by the goals of an interaction; the relational tasks, centered on the maintenance of the relationship, and the interaction management tasks, addressed to hold a coherent conversation. When establishing a relationship, the relational tasks are the priority, and consensus is the relational instrument used par excellence. As the relationship progresses into further stages (acquaintance, exploration and commitment), the relational work can be reduced, since the values of the relationship are already established and dissenting opinions no longer jeopardize them.
The type and strength of the relationship regulate a higher or lower use of mitigation devices in interaction. The stronger the relationship, the greater the possibility of encountering unmitigated language and vice versa, high degree of free opposition expression is expected in close relationships (Tannen 2002;Mandelbaum, 2003;Habib 2008;Choi and Schnurr 2014). Therefore, an intense use of mitigators by close partners may suggest specific exchange rules agreed upon.
Production, perception, degree of expectancy, impact and tolerance of opposing views in interaction are attached to the relational level, or as Canary (2003) indicates, individuals who are concerned [...] about maintaining relationships are more likely than others to engage in avoidance, whereas individuals who are primarily concerned about defending their personal identity are more likely to confront the partner (:527). Therefore, in resolving discrepancies, the preferred behavior is either to avoid confrontation, ignoring occasional dissent and centering attention on common views, or to negotiate the agreement. Estrangement is a minority response. In close relationships, disagreement may not have a big impact, since it may only affect the interaction, rather than the relationship, therefore individual opinions may prevail and confrontation is allowed. On the contrary, in non-established or weak relationships, open dissent and confrontation without relational work could end up in conflict.

Methodology
Our goal was to show that the expression of disagreement and the interlocutor's response were conditioned by the type and strength of the relationship held by the interactants. Therefore, for our study to show differences between relationships, the relationship itself had to be treated as a variable with different degrees of strength. For this, we decided that participants should interact in pairs, following the grounds of sociological theories such as Simmel's (1950) that states that the dyad is "the best focus for understanding social behavior" (Krackhardt and Handcock 2006:14). To obtain different degrees of strength in the relationships we controlled for the horizontal parameters (common history, degree of familiarity, socio-affective ties, quality of the relationship and interactional style), as well as the vertical parameters (status, power and social variables). To avoid additional variable effects, situational parameters such as the goal of the conversation (a relaxed discussion on current issues) remained constant.
In order to determine the strength of each relation, we ranked them on a scale according to the following parameters: 1. Social Distance. Taken as the degree of symmetry established in the relationship in terms of the status and power of the speakers. 2. Type of Relationship. Following the distinction proposed in Milardo (1988) and Milroy (2008), we distinguished two types of relationships, transactional and interactional. The first refers to relationships where members interact frequently but which purpose is merely transactional and they are not a source of material or symbolic support to each other. The latter are relations aimed primarily at relational exchanges, where members may or may not interact regularly but most importantly, provide direct support, advice, and criticism (Milroy 2008: 418). We considered that relationships moved along the continuum from the transactional to the interactional poles. The degree of proximity to each of these points determined the type of relationship. One could argue the existence of a central category in which both traits +interactional) met. In our view, this is an unlikely option, since the more interactional a relation becomes, the less transactional can be considered, that is, the personal closeness overcomes the transactional purpose of a relationship. According to this, we ranked each 3. Degree of Trust. Taken as the degree of intimacy members of the relationship. 4. Quality of the Relationship. Taken as 5. Interactional Style. Conditioned by the members, resulting in the prevalence of competition and differentiation, or trust and suppo All variables were weighted on a range of distance between the speakers and 4 to the selection of 7 pairs of speakers of Peninsular Spanish, age 25 below shows the classification of the sample. Relations ranging from 1 to 2.4 (pairs 3, 5, 6 and 7) were considered relationships, whereas relations between 2.5 and 4 (pairs 1, 2 and 4) were considered to be strong.
Data were collected through semi was presented with brief quotations from scientific works on male and fe stereotypes, such as lack of understanding between men and women (Tannen 1990;Gray 1992; among others). After the statement, the interviewer asked them to express their opinions about the content. All pairs discussed a total of eleven quotations. Conversations lasted between 60 to 90 minutes. All the data were collected by the same interviewer (In), who in all cases was known to the speakers. In order to naturalize the conversation, the interviewer held the role of participant each pair. Importantly, although our goal was to observe the use of language between the pairs of speakers, the study of their interaction had to be done by introducing a third person in the conversation. It has also been suggested that in order to study how Romera, M. CLAC 79 2019: 297-322 argue the existence of a central category in which both traits (+transactional In our view, this is an unlikely option, since the more interactional a relation becomes, the less transactional can be considered, that is, the personal closeness overcomes the transactional purpose of a relationship. According to this, we ranked each relation as 1[+transactional] 2[+transactional transactional +interactional] 4 [+interactional]. the degree of intimacy and familiarity between the Taken as how members get along (Alberts 1990). onditioned by the personality and the temperament of the prevalence of one of the two opposing forces of differentiation, or trust and support. on a range of 1-4, where 1 represented the highest and 4 the greatest closeness between them. This led the selection of 7 pairs of speakers of Peninsular Spanish, age 25-60. Table 1 e classification of the sample. Table 1. Sample of participants and degrees of strength in the relationships (pairs 3, 5, 6 and 7) were considered as weak relationships, whereas relations between 2.5 and 4 (pairs 1, 2 and 4) were Data were collected through semi-directed conversations. Each pair of speakers was presented with brief quotations from scientific works on male and female stereotypes, such as lack of understanding between men and women (Tannen 1990;Gray 1992; among others). After the statement, the interviewer asked them to express their opinions about the content. All pairs discussed a total of eleven quotations.
versations lasted between 60 to 90 minutes. All the data were collected by the same interviewer (In), who in all cases was known to the speakers. In order to naturalize the conversation, the interviewer held the role of participant-observer with Importantly, although our goal was to observe the use of language between the pairs of speakers, the study of their interaction had to be done by introducing a third person in the conversation. It has also been suggested that in order to study how the dyad behaves, it is best to place them in context, that is, in a group of three people (Simel 1908(Simel /1950. Indeed, several studies indicate that the behavior of a dyad is affected in three-person group situations, with the result of making the dyad's relationship even more cohesive (Mills 1957;Krackhardt and Handcock 2006, among others). That is, in three-person groups the dyad always takes primacy and the third member is taken as an "intruder", which makes the two-member relation even more cohesive. Cohesive behavior will make the speakers show their most prominent behavior as a pair, i.e. it would enhance both their agreement and/or their disagreement while interacting. The intruder can have different roles, from mediator in case of conflict, to the person who wins in it, but it seems that s/he always acts as a binder of the dyad's relationship. This made us think that the presence of the interviewer in the collection of our data did not disturb or modify the dyad's relation, and that it was even beneficial to better observe their behavior.
We are also aware of the problems of a semi-directed interaction, and acknowledge that it is not entirely comparable to a totally spontaneous interaction. However, in this case, an almost natural and spontaneous conversational style was achieved, since the topics turned out to be very successful in provoking discussion, and the familiarity of the speakers with the interviewer contributed to creating a relaxed situation. With the conversational settings unchanged, containment was only shown once by one speaker in an asymmetric relationship, which evidenced that the methodology used was not the main factor affecting spontaneity.
Finally, we are not claiming that the results of our study are unequivocally representative of the use of linguistic expressions of disagreement in Spanish according to the relationships hold between participants. A higher number of speakers would be necessary to prove correlations between the two.

Data Analysis
As we were going to analyze the resolution of dissent, for each conversation, we selected fragments containing instances of confronting opinions between the speakers. We coded them following our interpretation as analysts. Additionally, when parts of conversations proved hard to interpret, we returned to the interviewees and asked them about their communicative intentions. This information helped us obtain a more accurate interpretation and contrast our understanding with that of the speakers themselves. The data were analysed qualitatively.
Researchers in the field have proposed different disagreement taxonomies (Pomerantz 1984;Herrero Moreno 2000Brenes Peña 2011;Angouri and Locher 2012;Sifianou 2012;Albelda & Barros 2013). Mainly, all propose to classify the expressions of disagreement along a continuum according to the strength of the act's illocutionary force, ranging from strong disagreement with no mitigation to weak and mitigated disagreement. For Spanish disagreement acts Hernández Flores (1999), Herrero Moreno (2002) and Brenes Peña (2011) proposed similar taxonomies. Based on their proposals, for the present analysis two groups of linguistic expressions of disagreement were considered: expressions of strong disagreement and expressions of weak and mitigated disagreement. By strong disagreement we understood expressions by "which a conversant utters an evaluation which is directly contrastive with the prior evaluation' (Pomerantz, 1984: 74). Weak disagreements, on the other hand, were represented by the expression of partial agreements or partial disagreements (Pomerantz 1984:65)   Following this distinction, we now exemplify in greater detail the crucial role of relationships in interactional data. We highlight the fact that the expression and the impact of dissent emerge differently according to the type of relationship, and we also analyze how the disagreement resolves itself. We begin with the expression of strong disagreement, which occurs primarily in strong relationships. We then move to weak and mitigated disagreement which take place primarily in weak relationships, where disagreement represents a problem and settling consensus is a priority.

Strong disagreement
Our data shows that strong disagreement is allowed primarily in the strong relationships. The expression of disagreement with no mitigation can lead to a potential conflict between the interlocutors, however the strength of the bond between the partners, explains first, that the impact caused is much lower than expected and, second, when impact is caused, reaching consensus and maintaining the relationship prevails.
In the following example we see how the strong disagreement has no impact on the interlocutor, because the addressee chooses to ignore discrepancy.
Case 1. Absence of impact. Ignore discrepancy Pair 2 (C and D) is formed by two PhD students and lecturers in the same department as the interviewer (Int). Their common history goes back ten years. They have shared professional and personal experiences, and their way of thinking is similar. They both recognized that their mutual knowledge is considerable and the relationship is intense. They share a similar professional status and the relationship is based on similar terms of power. Their character traits are also similar; jokes and sarcastic comments are allowed in their rules of exchange, without either taking offence.
In the following excerpt, they talked about seduction techniques, in particular about nightlife and how seduction methods used nowadays by young people seem to have changed. C and D have conflicting points of view: while A defends that night bars do not offer a good context where one can meet interesting people, B acknowledges leisure time, entertainment and having fun as a way of forgetting problems.
(1) C.: el problema es que, cuando estás en un bar, dejas de ser quien eres tú para ser la chica del bar (. All: laughter The first expression of disagreement arises after C's consideration about the change that women experience when they are in a context of flirting. D responds with an expression of strong disagreement headed by the marker "pero a ver" / "but let's see". Her speech starts with a non-cooperative interruption, by which she gets to impose her turn over C, who is unable to finish her sentence. The expression used by D ("but let's see") points to a strategy of distancing. However, the subsequent inclusion of a counterargument, whereby D elaborates her opinion ("when you go to the cinema to watch a movie / imagine: you have a horrible day [...]"), redirects the distancing towards an approach, trying to make her friend converge.
This fragment is followed by a new reactive move from C, who begins a partial agreement. Her use of the expression "yes=, but then, that's why" indicates acceptance+opposition (Brenes Peña, 2011). But C follows her counterargumentation by raising her tone to regain her turn ("that's why I STOPPED GOING OUT TO BARS") to utter a supporting conclusion and make her point stronger.
After the interruption, C utters what can be interpreted as an accusation aimed directly at D through the personal pronoun "you": "because you go, but I don't go to have fun, I don't enjoy myself (.) sticking my tongue down some unknown's throat!". Realizing that the accusation can have a strong impact on C, B rapidly changes the use of the second person to the first person "I": "but I don't go to have fun, I don't enjoy myself". The attempt to go back on her words, changing to the first-person "I" mitigates any possible damage, and can be interpreted by D as a sign of repentance.
However, D doesn't seem bothered at all by the possible critical commentary, given her laughter and her collaborative sentence about Brad Pitt's presence, showing total agreement with A (there would have been slapping).
After the interview, we returned to the speakers and asked them about their perception of this fragment. A told us that her use of "you" was not aimed directly at D and D confirmed to have interpreted it in the same way. The coincidence in the interpretation responds to the high degree of knowledge they have of each other; the interviewer present at that time actually considered that "you" was directly aimed at B. That is, in the eyes of a stranger, C's intervention had been a clear accusation and only the speakers were able to interpret it differently.
Thus, this example shows how mutual understanding between partners and the exchange rules established enable the correct decoding of discourse. This explains why despite the different disagreement variants used by C and D, no conflict occurred in a strong relationship like this. Their interventions met their standards and their mutual expectations. Therefore, although some expressions might point to distancing, there was no real threat towards CD's relationship at any time, as was later confirmed by the speakers themselves.
Strong disagreement can also be solved by consensus, either by one of the speakers agreeing with the other or by both reaching the same conclusion. Case 2 shows an example in which after a clear impact produced by opposing opinions, one of the interlocutors chooses to avoid conflict by agreeing with the other. The conversation then resumes as usual and their relationship remains unaffected.

Case 2. Repairing impact by consensus
Pair 4 is formed by a mother (G) and a daughter (H). The mother, around 60, lives alone after recently separating from H's father. The daughter, in her thirties, is married and has young twins. H is writing her dissertation in the same department as the interviewer (Int) and they are friends. G also knows the interviewer. The status of participants in a parent-child relationship is generally considered asymmetric, since parents deserve respect throughout their lives, even when children have left the parents' home and/or have reached a higher social status than their parents. DelPrete (2015) analyses how disagreements without mitigation are characteristic of mother-daughter interactions given the close relationship holding between them.We could therefore say that in terms of categories, they hold an asymmetric relationship, although G and H have a strong bond and get along well as they meet nearly every day and share common interests (they are both teachers).
The mother-daughter conversation consistently shows strong disagreement which can be considered impolite. There are examples of insistence on one's personal point of view, correction and accusations. In this fragment, when the issue of men's preoccupation with their partners comes up, H tells the story of the letter her husband wrote to her to ask her out for the first time. This is something which her mother considers exceptional, as she does not think H's husband pays much attention to her in this way. The mother verbalises her opinion about the letter by saying "that's a lot for him!", as shown in the following example (2): (2) G.: Bueno, yo, aquí tengo que contar una anécdota, y es que M., no desplegó grandes artes de seducción PERO, cuando me pidió para salir, la primera vez, que éramos muy jovencitos, veinte años, me escribió una carta H.: ¡que eso para él es mucho! G: ¡¡Que eso para ÉL FUE=!!, bueno, ¡una gran hazaña! (2') G.: Well, I have to tell this story, M. was no expert in the art of seduction BUT when he asked me out, the first time, we were very young, twenty years old, he wrote me a letter. H.: That's a lot for him! G.: That for HIM WAS=!! well, a great achievement! Although this comment might have offended H, as it directly discredits H's husband, and therefore questions her choice to marry him, she does not take offence. On the contrary, she aligns with her mother -against her own husband, and emphatically supports her statement. She repeats her mother's sentence intensifying the tone and substituting "a lot" for "a great achievement", which strengthens the point made even further.
After a few exchanges, the mother restates her opinion of her son-in-law; as shown in (3).
( In.: @@ H explains that there were no more romantic gestures from her husband after that letter. G partially disagrees by uttering a sarcastic expression: "well, he once gave you a stone". The absurdity of giving a stone as a gift to a lover makes the mother and the interviewer laugh and reinforces the view that H's husband does not really engage in romantic gestures. The marker "bueno" (well) in this context acquires a clear concessive value, and is followed by a contribution that seeks to discredit G's son-in-law. The focus of the disagreement here is twofold: first, it is discrediting the false statement that H never received other romantic gestures by her husband, and second, it is a direct discredit of the alter, as it seeks to downplay H's husband and indirectly make fun of H in front of the interviewer.
At this point, the impact on H is clear, she does seem uncomfortable and ironically replies: "very funny!" She attempts to refute G's opinion by directly correcting G's statement and explaining that it was more than just a stone. However, G insists that it was no more than a stone he had found on an excursion, by an expression of strong disagreement accompanied by profuse signs of disbelief ("yes==!!] Oh Gosh! Because he got her a stone instead of a rose, plea==se! @@"). In the end, instead of disagreeing or showing offence, H decides to end the ascending conflicting spiral. She implicitly assumes G's criticism and justifies her husband's way by stating that "he is a very unpolished man of the country". The statement contributes to the collaborative construction of discourse and it represents a clear example of conflict resolved in favor of consensus and preservation of mother-daughter relationship.

Case 3. Exhausting disagreement without consensus
Finally, when strong disagreement occurs, strong relationships also permit partners to hold their own antagonistic opinions and not reach consensus. This was clearly the case of the interaction of the romantic couple.
A and B are a boyfriend and a girlfriend who are PhD students in the same program. Their relationship goes back five years, when they met and decided to move in together. They have a very close relationship, they get along very well, and have similar views about fundamental issues. However, this does not prevent them from contradicting or challenging each other's opinions almost continuously. Their style of disagreement is not aggressive or unpleasant, nonetheless. The subjects and the interviewer (In) were classmates in college. In the following example, B challenges A's affirmations about gender stereotypes, showing continuous unmitigated disagreement. The confrontation lasts over several turns, in which each interlocutor maintains their own position and no consensus is reached.
The interaction only ends when the topic is exhausted. In spite of the confrontation, no impact is shown in their relation.
The disagreement starts when B expresses her position regarding stereotypes, namely that they are more noticeable in adulthood. A opposes to that view with a clear "no" and the repetition of the adverb "precisely" to emphasize his disagreement.
(4) B: Cuando eres/ los niños que son más adolescentes, la edad así que son niños, nos no, la gente a veces suele decir, pues este tal, pero se nota menos porque son niños, […] In: Ahá A: No, precisamente van marcados desde pequeños, o sea, precisamente las páginas de catálogos de los juguetes marcan y dictan (4') B: When you are/ the children who are more adolescents, an age like that, who are children, we do not, people sometimes say, because this guy is such and such, but it is less noticeable because they are children […] In: U-huh. A: No, more precisely, they are marked since they are little, I mean, precisely the pages of catalogs of toys mark and dictate In her next intervention, B elaborates her position further, and A replies elaborating on his disagreement and confronting B's opinion overtly again. At this point B could have been offended by the confrontation, but she laughs after A's intervention.
(5) B: No buscan esos estereotipos hasta que no han llegado a la edad adulta […] A: [Yo creo que no...] B: Que hace lo mismo que cuando era adolescente y nadie le ha recriminado nada A: Yo no, no estoy de acuerdo, porque, precisamente, la niña que jugaba a fútbol es un machorro y el niño que está debajo de la escalera jugando a las lolly pocket es un mariquita. Tod@s: Risas (5') B: They don't seek those stereotypes until their adulthood […] A: [I don't think so...] B: they keep doing the same things as when they were teenagers and no one recriminated anything [to them]. A: I don't, I don't agree, because, precisely, the girl who played football is a tomboy and the boy who is under the ladder playing the lolly pockets is a fairy. All: laughs As we see, the confrontation is overt and clear in both of A's interventions. He uses the expressions "I don't think so" or "I don't agree", and he even interrupts B's turn, overlaps his words with hers and emphasizes his contrasting position using the adverb "precisely" twice. But it seems that A's disagreement is not offending B. Rather, A manages to elicit the laughter of both interlocutors by introducing in his argument the example of the small dolls (lolly pockets), with which both probably played.
The confrontation continues with several lengthy explanations by both A and B about their views on the subject. As it can be seen in (6), the interaction only finishes when it is obvious that both stand by their views and none of them gives way, at which point the interviewer suggests to move forward.
(6) B: Yo no digo que no marque pero a lo mejor mmm, los tópicos, ya, se buscan más en la edad adulta. A: I believe that we grow with the stereotypes and then they develop and, we mark a stereotype to define ourselves, or we do not mark it for.. or we hide it to take another one, because it is a stigma. In: Shall we move onto another (topic)? B and A: Yes, yes. After both interlocutors agree to move on to the next topic, the conversation goes on and they explain their point of view freely again, with no resentment from their previous confrontation.

Weak and mitigated disagreement
Unlike with strong relationships, strong disagreement is almost absent in weak relationships, the search for consensus is a priority and primarily weak or mitigated disagreement is found. Since dissent can easily cause a negative impact on the relationship, it must be expressed carefully. Usually, speakers disagree in an indirect manner and address their dissent towards the content rather than to their interlocutor. Some of the formulas used are questions, 1st person plurals and impersonal expressions which help depersonalize the discourse and avoid confrontation and detachment from the other. This way the relationship is not damaged and no repair work is necessary. The end of these interactions is always to reach agreement, through mutual convergence or by having the other person converge.

Case 4. Converge
The following interaction occurs between members of pair 3. E and F hold a symmetric relationship; they are two male cousins between the ages of 20 and 30. Both maintain a cordial relationship, but do not share other experiences beyond very occasional family gatherings. The interviewer (In) is also a relative of theirs, so all three know each other and feel comfortable in the conversation. They have been talking about whether the stereotype of the "kept woman" still remains in today's society. Disagreement is expressed indirectly, F avoids dissenting in a direct manner and does so in the form of a question requesting clarification at the same time that he justifies his objection through an exemplary case. E on the other hand, avoids conflict and immediately converges to the position of F. D: that is the, the, the= well, that, the, the, the minx, the clever one E: [yes F: that] is looking for that, and then the other women who existed especially in past. In this excerpt when the interviewer asks about whether E and F believe that there are women who live off their husbands, only E responds, and he does it emphatically, stating that this image "exists". F's silence makes the interviewer ask for confirmation a second time, and once more it is only E who responds, more emphatically this time ("and there are many! many!"). At this point, F decides to show his discomfort with the issue. However, as already said, he expresses his disagreement indirectly in the form of a question, and without addressing any of the speakers, particularly E, who has already shown his agreement. F's turn, initiated with "but", only calls for specification of the content of the conversation, specifically on what kind of women the word "kept" refers to. In addition, F uses the 1st person plural "we mean" to avoid personalizing.
After this, F elaborates his question, justifying his disagreement with the fact that all women who live off their husbands should be classified alike. The common knowledge shared by the participants, -all of them relatives, indicates that F may have felt offended, because his mother is a woman who has always stayed at home taking care of her children and never worked elsewhere. The impact has been caused indirectly due to a lack of consideration for F's circumstances.
Inadvertently, E has not taken into account the situation of F's mother when answering, and his emphatic statement may have indirectly resulted offensive to F. E immediately realizes his error and deactivates a potential conflict. Unwilling to wait for F to finish his turn, E emphasizes his agreement with F in two collaborative interruptions, "I agree with him" and "of course" which affiliates him with F and ends the controversy.
As we can see, in this 'weak' relationship the impact has occurred soon and indirectly, only after an inference made by one of the interlocutors. F thought that by his emphatic statements, E was not taking into consideration his circumstances; otherwise E's answer should have been different. E reacts to this converging immediately and disabling the potential conflict automatically.

Case 5. Make the other converge
The next interaction in (8) shows the use of cooperative strategies when expressing dissent in asymmetric weak relationships. Both speakers (KL) are very careful not to offend each other when disagreeing, and both seek consensus. The boss (K) is in his sixties and runs a medium sized business. He recently hired L as a worker. They did not know each other previously and their relationship is basically transactional, which means they lack particular standards of behavior agreed upon by them, and both behave according to their status. The interviewer (In) knows them both; In and L were classmates in college and In and K are close relatives. During the conversation, K has expressed his view about the superiority of women throughout the history, now and in the past. L, on the contrary, believes that women did not have the opportunity to develop their full potential until recently, and that gender stereotypes still remain. In his previous intervention, K claimed that he cries if a film moves him and he does not interpret this as a sign of weakness. L then expresses weak disagreement, via partial disagreement, mentioning that there are still people who believe that men should not cry. K's disagreement is again introduced in an indirect manner, through a personal discourse, in which the use of first person forms and verbs of opinion dilute the potential impact on the interlocutor. Thus, after several hesitation expressions (to me eh= / I, well), L introduces a disclaimer on the opinion he is about to issue "everyone is different". The same strategy is used again when "I see it" is used to counter the opinion of K, and also when "I think" closes the first unit and opens the second part of L's discourse showing his contrary opinion to what K said. To further minimize his discrepancy, L includes the acknowledgment of K's views in his discourse. This recognition serves as a positive minimizing element prior to the expression of disagreement. L begins his intervention recognizing K's argument as "true". And later he includes K's argument "not only by physical force", aligning it to his own argument, instead of arguing against it.
K, on the other hand, uses expressions of overt disagreement, as we can see when he utters a plain negation "no" in response to L's statements. K also personalizes his disagreement pointing directly to the other when he begins his argument with an expression addressed directly to L, "you are coming to my field", treating him as his opponent. However, K immediately minimizes these confrontational strategies with the expression of cooperative disagreement. K incorporates L's arguments to his own discourse, accepting them and granting them the same value. K completes L's intervention when he argues that "women are more capable", paraphrasing L's words as "women surpass them". Also K incorporates L's argument about "the lack of security of men who beat women" and he equates it to the one he previously put forward, "men feel inferior." K combines both arguments as to reach consensus.
(8) L: Claro, pero sí según quien, interpreta eso como todavía como muestra de debilidad/ a mí eh=/ yo, bueno, cada uno es como es/ pero viendo cómo se (..)/ viendo las noticias cómo crecen/ cómo suben los/ los/ las nuevas generaciones, viendo que el número/ el número de/ de violencia machista no desciende, el número de asesinatos, el número de maltratos es que sigue/ siguen funcionado determinados estereotipos K: Vienes/] vienes a mi terreno. (..) ¿Por qué se da la violencia/ la violencia de género mucho más del hombre a la mujer, que de la mujer al hombre que también existe? Por la FUERZA FÍSICA, que es en lo único que podemos plantar batalla, ¡en lo otro es que no tenemos nada qué hace=r! In: @ L: Cierto, pero yo lo veo motivado por la falta de seguridad. K: No/ L: El hombre que agrede a la mujer es porque tiene falta de seguridad en sí mismo In: Eso]/ eso dicen los [terapeutas L: Y se sienten]/ y se sienten amenazados precisamente por una/ por una mujer que, por lo que se ve que es [más capaz que/ K: que les ga=na]. Es que, va=mo=s L: Pero no sólo por la fuerza física creo, creo que es por la falta de seguridad K: Bien, pero es que una cosa es consecuencia de la otra, es decir, estamos hablando insisto en términos de promedio (..) ¿Por qué tienes que/ que recurrir a esta cuestión? Pues porque/ porque eres inferior, por falta de seguridad en ti mismo, por falta de poderte sentir realizado, porque el hombre ha sido quien/ a quien la/ para mí la mujer es TAN INTELIGENTE ¡que es que le ha cedido el papel que ha querido!, (..) (8') L: Of course, but depending on the person, he interprets that still a sign of weakness / to me eh = / me, well, everyone is different / but watching how (..) / watching the news how they grow up / how new generations rise, seeing that the cases / cases of / violence sexist [violence] do not descend, the number of murders, the number of cases of abuse is that it continues / certain stereotypes still work K: you're coming to my land (...) Why does violence / gender violence occur much more from men to women, than from women to men, which also exist? Because of PHYSICAL STRENGTH, which is the only thing we can plant battle, with other things we have nothing to do=! In: @ L: True, but I see that it is due to the [man's] lack of security. K: No/ L: The man who assaults a woman is because he lacks self-confidence In: that's] / that's what therapists say L: and they feel] / and they feel threatened precisely by a / by a woman who to them to them is [more capable than / K: she wins them=] It is that, of course= L: but it's not only because of physical strength I believe, I think it is because of their lack of security K: well, but one thing is a consequence of the other, that is, we are talking, I insist on average terms (..) Why do you have to / to resort to this question? Because/ because you are inferior, for lack of security in yourself, for lack of being able to feel fulfilled, because the man has been who / to me the woman is SO INTELLIGENT that it is that she gave him the role he wanted! (..) In this example we see how the different status of K and L in this relationship allows for K to disagree, while it is not the case for L. The absence of personal exchange rules between K and L, i.e. specific terms for their relationship agreed upon by them, causes both partners to follow the norms established for this type of relationship (boss-employee) in their community (Marra 2012). However, precisely because of the need to preserve the relationship, neither, especially L, expresses disagreement confrontationally. Rather, both try very carefully not to impact the other. Due to his status, K can take issue more openly and he even personalizes his disagreement. However, he always mitigates his disagreement immediately and tries to reach consensus, incorporating L's arguments to their own and developing a joint argumentation to provide validity to both views.

Conclusions
We began this article with the objective of showing that social relationships are regulatory mechanisms of linguistic behavior in interaction that have a much more important effect on our discourse than put forward until now. Each relationship brings into play a set of circumstances common to its members, which regulates how they behave and how they interpret each other's words and actions. It creates a 'relational culture' (Burleson, Metts and Kirch, 2000). Thus, we have shown that the preservation of relationships acquires a prominent dimension in our interactions.
A relationship is a dynamic element which changes with daily interactions (Arundale 2010). Our behavior in interaction cannot be explained only in terms of individual behavior, since this would not explain why when individual conditions are seemingly identical (status, power, etc.) people show a different linguistic behavior. As we have seen in our analysis, the terms agreed upon in the relationship or the "exchange rules" are crucial for the interpretation and impact of linguistic expressions.
We have shown that in the expression of disagreement, the impact of fairly shocking strategies of dissent, such as correction, accusation, discredit or overt disagreement depends on the standards agreed upon by the speakers according to the strength of their relationship. This strength determines the greater or lesser impact of opposing views and the higher or lower use of mitigation devices. The stronger the relationship, the lesser the impact and the lower the need for mitigation. In contrast, when relationships are weak, disagreement can easily generate a negative impact so the use of mitigators must increase to intensify the relational work.
In almost all the examples examined, the basic principle guiding interaction is to reach a consensus among speakers, which in turn strengthens the bonds and solidarity between them and maintains the relationship healthy. The consensus can be achieved through various strategies, such as trying to make the other converge towards our point of view, converge ourselves to the point of view of the other, or even ignore the impact of a direct expression of discredit and guide the conversation towards topics where agreement can be reached more easily. In the same way, if an expression allows for multiple interpretations, speakers will choose the most appropriate depending on the strength of their relationship. Thus, in strong relationships, participants guide their interpretations towards the most favorable ones in order to preserve their relationship. It is only in strong relationships that we observed also cases in which holding each other's position is possible without impacting the relationship. In this case, the principle of reaching consensus is overridden by the power of the relation.
From our results, we can conclude that there is a significant link between the strength of the relationship between the interlocutors and the regulation of disagreement in interaction. These results also point to the general importance of social relationships in discourse. Social relationships affect the basis of communication between individuals. The recognition of the affiliative principle, which seeks to maintain social relationships in the best possible terms (Cf. Enfield 2006;, as regulatory tool of discourse, opens the door to new ways of analyzing interaction.