Mitigation Strategies in Colloquial Conversations: A Contrastive Approach in European, Cuban, Mexican, Argentinian and Chilean Spanish

This paper compares the pragmatic mitigation strategies used by speakers of five geographic varieties of Spanish: European, Cuban, Mexican, Argentinian and Chilean Spanish. Mitigation can be understood as a pragmatic strategy arising from face-pro-tection needs that seeks to reduce possible adverse effects on the satisfactory progress of communication. The data are drawn from the Ameresco corpus of colloquial conversations, and the analysis includes four kinds of parameters: linguistic, pragmatic, and socio-situational. The study explores the functions and frequency of use of mitigation, speakers’ and hearers’ exposure to social and relational risks, the degree of their commitment in interaction, and the most productive resources observed in each variety. The results show convergences and divergences in the use of mitigation in these geographical areas, providing a picture of dialectal and sociolectal patterns here. The study concludes that mitigation varies according to the priorities of face needs in each speech community.


Introduction
Recent decades have seen exponential growth in research looking at pragmatic mitigation and how it varies geographically.In particular, work within the PRESEEA1 project is currently contributing to the compilation of a geolectal and sociolectal map of the uses of mitigation in Spanish-speaking areas (Albelda and Cestero 2011, Cestero 2017, Samper 2017, Albelda et al. 2021, Guerrero 2021, Cestero and Albelda 2023).However, this research is based exclusively on semiformal interviews.Indeed, dialectal variation in spontaneous, informal conversation has received little attention to date.As a mode of language and a discursive genre, colloquial conversation is the commonest and most pragmatic form of human communication (Givon 1979).Thus, the use of any linguistic mechanism (in this study, mitigating strategies) in this discursive genre can be seen as emerging in an absolutely free and unrestricted way, as opposed to language produced in an interview format, for example, which is likely to be constrained by a number of situational and social factors (Givon 1979, Biber 1988, Albelda 2008).
The current paper seeks to compare and contrast pragmatic mitigation strategies in colloquial conversations, as used by speakers of five geographical varieties of Spanish: European, Cuban, Mexican, Argentinian and Chilean Spanish.The following research questions will be addressed: RQ1.What are the quantitative and qualitative similarities and differences in mitigation strategies used by speakers in European, Cuban, Mexican, Argentinian and Chilean Spanish in spontaneous conversations?Are there any differences observed here between women and men?RQ2.Taking mitigation as a pragmatic strategy arising from the needs of face-protection, what are the degrees of speaker and hearer exposure to face, and the degree of speaker commitment here in each geographical variety?
The study is based on an analysis of 500 minutes of oral speech, 100 minutes of each variety/dialect, this material largely drawn from a section of the Ameresco corpus (América Español Coloquial, Albelda and Estellés online).According to these currently available data, and given that the study is a first descriptive approach to geographical variation in mitigation that focuses on spontaneous Spanish conversations, some limitations have been used in data selection relating to sociolectal features.Specifically, the data pertain exclusively to younger speakers, with a total of 78 subjects aged between 18 and 30, these all having a high (university) level of education.Subjects are divided equally in terms of sex (39 women and 39 men).
The Ameresco corpus is comprised of spontaneous, everyday conversations recorded in natural settings.They are informal in the sense that they are unguided, hence neither the number of speakers involved, the topics covered, nor the lengths of the conversations are predetermined.Also, during the conversations themselves speakers are unaware that they are being recorded (see Section 3.1).This is of particular value for the present study.Previous work on pragmatic (and therefore contextual) phenomena such as mitigation has typically been based on elicited methods; the present approach, by contrast, deals with naturally-occurring data which, as such, are wholly uncontrolled by the researcher (see Félix-Brasdefer and Hasler-Barker 2017, Koike 2021);2 no elicitation cues were given to speakers, and hence their speech production is wholly spontaneous.As instances of freely emerging talk, then, the data might include dialectal variation, in that a group of between 14 and 18 speakers was recorded for each variety.
As we will discuss below, mitigation is a strategy related to social face that varies markedly across the spectrum of formal to informal communicative situations (De Cock et al. 2018, Albelda 2018, Figueras 2021).In formal contexts, such as interviews, uses of mitigation tend to arise largely due to the distance between interlocutors, the transactional purposes of the interaction, and the conventionalisation of certain communicative activities.In informal situations, by contrast, the relationship between speakers is likely to be far closer, the context private and the communicative purpose interpersonal, and hence the individual can "relax, take off the mask, leave his/her text and leave his/ her character aside" (Goffman 1959).Leaving one's mask aside does not imply that one's face is wholly unprotected; rather one's face changes, becoming more sincere and authentic in nature, what we might call a face without the mask (Bravo 1999, Albelda 2008, Hernández-Flores 2013).Informal, spontaneous contexts allow us to observe when speakers naturally consider the use of 2 As Koike (2021: 568) pointed out, "elicited data entail methods through means that would not occur by themselves".In contrast, spontaneous conversations will still occur even if they are not recorded.In the tradition of Val.Es.Co.Group studies (www.valesco.es),corpora of colloquial conversations are conducive to authentic and real analyses of the most common and frequent discursive genre, because they occur in real and natural contexts.mitigation appropriate and well-motivated, with no previous, a priori determination or imposition of the kinds of communicative activities in which we believe mitigation might arise.
It should be noted that in what follows a broad perspective on mitigation has been adopted, in that this is an initial approach to the study of dialectal variation in these conversational data, and future work is expected to continue and expand on the findings.The unconstrained nature of these data makes it more difficult to compare samples, but the first steps here are intended to establish an analytical approach that can be broadened in scope for future studies.
In Section 2, below, a review of the concept of pragmatic mitigation in light of previous research is offered.The data and methodology are then described in Section 3. The quantitative and qualitative contrastive results of the analysis, plus a discussion of these, are presented in Section 4, before some concluding remarks on the similarities and differences in speakers from the five Spanish varieties are given in the final section.

2
Theoretical Background

Origins and Development of the Pragmatic Phenomenon of Mitigation
Mitigation can be understood as a pragmatic, strategic phenomenon in social communication.It has been studied in great depth over recent decades,3 and a solid theoretical base has thus emerged over the years.As a pragmatic phenomenon, mitigation has been characterised in a number of ways.Lakoff's study of hedges (1972), for example, looks at those expressions used to reduce the value of what is said or to increase or reduce the fuzziness of meanings, whereas other models have described how the illocutionary force of utterances can be downgrading in some way by speakers (Fraser 1980, 2010, Holmes 1984, Caffi 1999, 2007, Kaltënbock et al., eds., 2010, Briz 2007, Czerwionka 2012, 2021, and many more).
As Caffi (2010) has pointed out, reference to mitigation as a phenomenon of rhetorical-argumentative persuasion can be found in Rhetoric for Herennius (of unknown authorship), the oldest rhetorical treatise in Latin.Some linguistic and grammatical treatises also refer to semantic-linguistic graduality and scalarity (see for instance, Bello 1847, Bolinger 1972, Labov 1984, Quirk et al. 1985).However, modern study of the phenomenon only truly emerged with the advent of study of modal logic, especially in the research of Zadeh (1965) and Lakoff (1972).
In modal semantics, the framework of fuzzy concepts posits that, in addition to the two polar truth values (true and false), other intermediate degrees of truth can be expressed.These are called hedges, expressions such as sort of, like or somewhat, and are used to reduce the truth value of an utterance or to present reality in a vague or imprecise way.
The development of fuzzy logic emerged partly from a pragmatic understanding of such linguistic usage, with its communicative effect on language seen as effectively reducing the speaker's commitment to what is being said and introducing an intentional vagueness of expression.A great deal of research here has focused on mitigation as a pragmatic category that downgrades the illocutionary force of a speech act (Fraser 1980, Holmes 1984, Bazzanella et al. 1991, Sbisà 2001, Thaler 2012).In some of these studies the various interactional dimensions on which mitigation can have an impact are established: propositional attitudes, inner states, the modal roles of participants (speaker's commitment, addressee's obligations, etc.) and perlocutionary goals (Holmes 1984, Bazzanella et al. 1991, Caffi 1999, 2007, Briz 2007).
Many such studies also highlight the close relationship of mitigation to social aspects of communication, in that mitigation can be used in speech acts which might have an uncomfortable or hurtful impact on the hearer's face.A prototypical case is that of directive speech acts, through which the hearer is urged to do something which may require great effort; mitigation is used in here to care for and protect the hearer's image or territory against possible harm or perceived threat (Fraser 1980, 2010, Haverkate 1988, Garcés-Conejos and Torreblanca-López 1997, Márquez Reiter 2000, Cheng 2001, Hernández-Flores 2013, Albelda 2016, Flores-Ferrán 2020).Mitigation has also received considerable attention in sociopragmatic and sociocultural studies of verbal politeness (House and Kasper 1981, Brown and Levinson 1987, Haverkate 1988, Locher and Watts 2005, Félix-Brasdefer 2008, Bravo 2021, among others).However, mitigation and politeness should not be confused, for two reasons.First, it has been argued widely that mitigation is a linguistic (pragmatic) phenomenon, whereas politeness is a social-behavioural one (Haverkate 1988, Briz 2007, Kaltenböck et al. 2010, Hernández-Flores 2013, Cestero 2020, Figueras 2021, among others).Second, not all mitigating strategies are intended to express verbal politeness; only when this strategy aims to protect or save the face of the hearer can it be considered a polite use, since its purpose is thus social (to prevent risks to a hearer's face: "Could you help me with the suitcase, please?").There are many examples of mitigation used to protect the speaker's self-image (Cheng 2001, Bernal 2007, Albelda 2016, Figueras 2021) which do not arise from a desire to save the face of the other interlocutor.Thus, in the example "One's strength isn't what it used to be" the speaker is referring to herself.

2.2
Characterisation of the Pragmatic Phenomenon of Mitigation As we can see in the literature on mitigation over the last three decades, most cases exhibit the same characteristic features: (i) mitigation is considered a pragmatic strategy, (ii) it arises from the needs of face-protection, (iii) it seeks to reduce possible adverse effects on the satisfactory progress of communication, (iv) it is a strategy used in the interpersonal negotiation of communicative agreement and of persuasion, and as such (v) it allows speakers to establish a lower level of commitment to what is being said, and hence to attain conversational goals more efficiently.
The above characteristics highlight the rhetorical and social nature of communication.4However, a cognitive approach to the study of mitigation has also been developed, albeit less extensively.Martinovski (2006Martinovski ( : 2066)), for example, proposes that mitigation arises when the speaker cognitively foresees "(a) dis-preferred outcome/s of a future action and offers a mitigated version of that action before it happens".Hence, mitigation occurs mainly in cases where a speaker fears that their self-image might be endangered because some utterance or other action potentially harmful to themself is going to be said or performed (in this sense, see also, Figueras 2021).
In the present study we will adopt a definition of pragmatic mitigation that combines the diverse perspectives encompassed in the literature reviewed above, but based most closely on Albelda and Estellés' (2021) proposal, which will be described below.We believe that a broader characterisation of the phenomenon will be helpful in distinguishing between uses which are genuinely mitigating from those which, in their context, are not.Our definition, then, characterises mitigation in terms of three perspectives, these being complementary rather than mutually exclusive: cognitive (the nature of attenuation), socio-rhetorical (purpose in communication), and linguistic (linguistic-rhetorical resources; Albelda and Estellés 2021).We will now examine each of these separately.
4 See also discussions on the characterisation of mitigation in Briz (2007), Caffi (2007) and Albelda and Estellés (2021).From a cognitive perspective, mitigation is defined as a linguistic strategy aimed at maintaining the assumptions that a speaker attributes to a hearer, these assumptions being related to the speaker's self-image.The speaker begins from the position of accepting these assumptions that are attributed to the hearer, and the purpose of the mitigation is to avoid such assumptions changing for the worse (see Albelda and Estellés 2021: 74-77).
From a socio-rhetorical point of view, mitigation is characterised as being used for social reasons in human communication and for the purposes of rhetorical persuasion.It serves here to prevent or repair possible undesired perlocutionary effects on the interlocutor's face in a specific communicative action, such as to soften a request made to them, or to mitigate self-blame on the part of the speaker.So, mitigation involves protection of the social face of the interlocutors, either of the speaker, the hearer, or both (Hernández Flores 2013, Albelda 2016).These social and rhetorical aims are inseparable in that social management is a basic requirement for the achievement of the aims of communication.
From a linguistic perspective, mitigation arises from an implicature, which may be more or less context-(in)dependent, that is, more or less conventional.Mitigation can be interpreted according to Levinson's (2000) heuristics of manner, in which a speaker achieves "a non-prototypical reading by a marked expression".Such mitigation leads the hearer to interpret a certain linguistic device as an insufficient, incomplete or vague form, rather than taking it to be information that they know to be true (Albelda andEstellés 2021, Cestero andAlbelda 2023).
In the application of these three broad criteria, it is important to bear in mind that mitigation is not linguistically coded.It is, rather, a contextual phenomenon in which there is no visible correlation between function and linguistic form.Hence, as well as formulating an exhaustive definition based on these different perspectives, it will also be necessary to carry out an analysis of the data which incorporates contextual information, as described in Section 3.2, below.
Numerous classifications and taxonomies of mitigating linguistic devices have been proposed, not least because languages tend to develop new forms and devices over time (see the various classifications in House and Kasper 1981, Prince et al. 1982, Hübler 1983, Bazzanella et al. 1991, Markkanen and Schröder 1997, Hyland 1998, Briz 2007, Caffi 1999, 2007, Albelda and Cestero 2011, Czerwionka 2012, Albelda et al. 2014, Johansen 2021, among others).A variety of linguistic devices can be used to express mitigation: phonic (e.g., suspended structures), morphological (e.g., diminutive suffixes), syntactic (e.g., downgrading quantifiers), lexical (e.g., softeners) or modal (e.g., question tags).1163/26660393-bja10087 | Contrastive PragmaticS (2023) 1-34 Among the broadest taxonomies are Caffi (1999Caffi ( , 2007) ) and Briz (2007), Albelda et al. (2014), the latter two relating to Spanish.In fact all three of these studies share the idea of distinguishing three mitigating categories or types, according to the three components of the utterance in which the mitigation can function.The first of these is a group of strategies that affect the propositional content (litotes, euphemisms or devices used to blur the meaning: "like", "kind of").Caffi (2007) calls these bushes, and Briz (2007) refers to them as dictum mitigators.Second is a group of strategies that influence the illocutionary force, reducing the speaker's commitment to the utterance (e.g., epistemic disclaimers: "If I'm not wrong").Caffi (1999Caffi ( , 2007) calls these hedges, whereas Briz (2007) prefers the term modus mitigators.Third is the group of strategies that mitigate the origin or source of the utterance, called shields by Caffi (1999Caffi ( , 2007)), whereas Briz (2007) considers them to be pragmatic mitigators of the participants' roles in the utterance (e.g., the use of we or one instead of 'I' or 'you').

3
Data and Method

3.1
Corpus Data There are very few corpora of spontaneous colloquial conversation for different geographical varieties of Spanish.For the present study, most of the data have been drawn the Ameresco corpus (América Español Coloquial, Albelda and Estellés online).This is a freely-available macro corpus, with direct internet to the complete conversations in audio format and transcriptions of these.More samples are currently being added to the database, which presently covers a number of geographical varieties of Spanish, including the five to be addressed in this research.
The conversations are relaxed and spontaneous, taken from exchanges between family members and friends.They were recorded in natural, everyday settings (homes, leisure areas, etc.) in normal, non-transactional socialising activities, between 2017 and 2022.Typical themes include what people like and dislike, pastimes, leisure, health, friends, future plans, personal and common problems, and so on.The Ameresco corpus was designed and compiled according to the following conversational and contextual criteria (Sacks et al. 1974, Briz and Val.Es.Co. Group 2002, 2014): (i) situational features: relationships of solidarity, social and functional equality between speakers, mutual relations of social and functional equality, interpersonal (socialising) purpose, lack of planning; (ii) conversational features: face-to-face interaction, Downloaded from Brill.com 12/28/2023 09:15:18PM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/no predetermined turn-taking, immediate mode of action, cooperative and dynamic progression and distribution of communicative roles.
The recordings were made in accordance with all the relevant data protection requirements (Organic Law 3/2018, 5 December, Spain) and under the supervision of the University of Valencia Ethics Committee.Although speakers signed their written consent in the weeks previous to recordings, they were not usually aware of being recorded at the time,5 and were informed of this subsequently, and also reminded that they could revoke their permission at any time.
The sample is based on the analysis of 78 speakers from five Spanishspeaking countries: Argentina (14 speakers), Mexico ( 14), Cuba ( 16), Chile ( 16), and Spain (18).Although the number of speakers in each group is not constant, 100 minutes of conversation are analysed in each case.These differences in the number of speakers are inevitable in that the conversations were spontaneous, undirected, and otherwise unrestricted, and hence no a priori prediction of the number of speakers could be made.
The study also seeks to observe possible gender differences in the use of mitigation, and thus is gender-balanced, with 39 male speakers and the same number of females.Furthermore, these speakers are all young adults, between 18 and 30 years old, with a high level of education (i.e., speakers were either university students or graduates).
We sought to make sure that the majority of data pertaining to each variety involved speakers from the same city, although in the cases of Spain and Chile this was not possible, due to size of the dataset.Thus, for Chile the data are from conversations recorded in Santiago de Chile (Ameresco corpus, González) and Temuco (Ameresco corpus, Mondaca), and for Spain the conversations were from Valencia (Ameresco corpus, Albelda and Estellés) and Granada (España, COGILA corpus, Barros et al. 2012).The total of 500 minutes of recordings is distributed as follows6:

Parameters of Analysis and Classification of Mitigation Strategies
In order to identify a speech act that is affected by mitigation, the three criteria discussed in Section 2.2 have been applied: cognitive, socio-rhetorical and linguistic.These allow us to see whether a specific device carries a mitigating function at a specific point in the discourse.In the following example (1), from Buenos Aires, we might well ask whether the form creo (I think) functions as mitigation in the context.Here two couples having dinner together in the home of one of them.The hosts, C and D, ask A, one of the guests, what he wants to drink.
( Let us examine the criteria discussed in Section 2.2 in this sample.From a cognitive perspective, Speaker A wishes to keep (not change) the positive image that his friends have of him; although he clearly wants to drink beer, he uses creo in order not to impose his preference but rather to express this as a probability.In fact, after saying creo que cerveza ('beer, I think') he corrects himself, showing his willingness to drink wine if the others prefer it.From a sociorhetorical perspective, we can observe his calculation of the possible harm to the hearers' face (C and D are serving the drinks), and to his own self-image (he does not want to damage his image as a good guest in front of his friends) and hence he expresses the fact that he is not fussy, that he can adapt, etc.Finally, from a linguistic perspective, the mitigation arises in a heuristic way.Creo is a marked use in this case, since the speaker is presumed to know what he wants to drink (he himself is the source of that knowledge), and he therefore does not need to express it using a verb of belief, since this (by default) would suggest that he himself were not certain of his preference here.
The opposite occurs in the following sample (2), from Havana, in terms of the same three mitigation criteria.Speakers A and B are talking about a television soap opera.B answers A's questions about what has happened, since A has given up watching it.Let us focus here on the use of no sé ('I don't know') as an honest way of expressing ignorance to A's question, without expressing any mitigating value: (2) B: ¡ay! ahora no me acuerdo qué pasa con Vega A: ¿y qué sabes del que quería ser escritor?B: ¡ah! ¿Thales?/ tampoco lo sé/ me dijeron que se queda con una chiquita que es prácticamente idéntica a Nicole/ que es prima o algo de Nicole/ no sé/ algo de eso me dijeron/ no sé Given the context of the conversation fragment in (2), the use of no sé ('I don't know') does not appear to be motivated by reasons of face protection of any of the speakers, since there is no suggestion of a potential threat.In this regard, neither the cognitive nor the socio-rhetorical criteria would be applicable.Similarly, no sé in this context is not a marked use, but a prototypical one; it is the expected formulation to express ignorance or uncertainty.
For our present purposes we have compiled a list of 20 strategies (linguistic and paralinguistic resources) which are commonly used in mitigation.This list is based on existing taxonomies of linguistic devices typically used in mitigation in Spanish (see 2.2, above), especially Albelda et al. (2014), Cestero (2020), and Cestero and Albelda (2023).Clearly, these are not the only possible devices available, and thus we also include an open variable in the analysis, in order to register those strategies that are not included in our initial list.
These 20 resources have been classified into seven broad groups according to the general strategic and communicative moves that are activated in negotiation and that approach the communication from the I-speaker's stance from less to more involvement (categories I and II), from the communicative content itself (categories III, IV and V), and from the You-interlocutor's position in terms of less to more involvement (categories VI and VII) (Cestero, 2020).These are presented below, but for more detail see Briz (2007), Albelda and Cestero (2011, 2013, 2014) and Cestero (2020).

I.
Resources It should be noted that the strategies listed here are not mutually exclusive and that several may be used in a single speech act.Hence two decisions have been made in the analysis.First, the total number of strategies per speech act has been taken into account.We will see in the results (Section 4) that in some cases as many as seven devices are used to mitigate a single speech act.Also, two separate counts have to be made for each variety: the number of speech acts mitigated, and the number of mitigating strategies used.See the following sample (3), from Mexican speakers, in which one mitigated speech act can be observed, but with six mitigating strategies in this single act: (3) C: hay ideologías muy arraigadas// de repente uno se siente ofendido y la gente ya no se habla// por ejemplo/ la familia de mis jefes.D: y lo malo es cuando eso afecta a la descendencia ¿no? y te dicen ya no le hables a tu primo// ¡wow! te ponen en una situación comprometedora/ se fractura toda la familia por cosas bien bobas A: bueno es que a lo mejor nosotros sentimos que son cosas bobas/ pero puede ser que sea algo que ya estén arrastrando desde antes de que nosotros existiéramos ¿no?
R. Maldonado, Corpus Ameresco-CdMexico.Conversation MEX-14 [C: Some ideologies are very deep-rooted.Suddenly someone feels offended and people stop speaking to each other, like my bosses' family.D: And the worst thing is that this affects the children, doesn't it?They tell you not to speak to your cousin any more.Wow!They put you in a compromising situation.The whole family is broken up because of these stupid things.A: Well, it's just that maybe it's us that think they're stupid things, but perhaps it could be something that's been nagging at them since before we were born, right?] Speaker A disagrees with what D has previously said, and he mitigates this with the following strategies: he provides a concession by using well; he opens his explanation with the justifier it's just that; he uses an adverb of probability, maybe, to leave the door open for other possible opinions; he uses a concessive structure with but, followed by a verbal structure, again expressing probability; and he concludes with a question tag, ¿no? ('right?').Since mitigation is a contextual phenomenon, recognising each specific use can be made easier by examining those socio-situational factors which, according to previous studies, might affect its use (Albelda 2013, Albelda et al. 2014, Cestero 2017, 2020).According to these authors, face is the key element in defining mitigation functions: mitigation can prevent potential damage to the speaker's own face (self-image) or can aim to avoid damage to the faces of hearers (hetero-image, sometimes considered to be politeness).Also, in the present study we have analysed the following parameters for each act which appears to be mitigated: the type of illocutionary force, the topic of the conversational fragment, the (un)equal functional and social relationship between the interlocutors, and the level of proximity and familiarity between speakers.

Comparison of Frequency of Mitigation Strategies in Each
Geographical Variety This first series of results focuses on the comparison of the quantitative data in each variety sample.Graph 1 shows the overall results of the analysis of the speakers from the five groups in ascending order of use of mitigators.The number of mitigated speech acts per 100 minutes in each geographical variety is shown, plus the total number of mitigation strategies used in each.Not surprisingly, there are a greater number of strategies than speech acts, since many speech acts contain more than one mitigating strategy.
As can be seen in Graph 1, the number of acts of mitigation used by the speakers of European Spanish is the highest of all five groups.In terms of the number of mitigation strategies used, the results from Spain are far ahead of those from the four Latin American varieties.For the number of speech acts, the results for European Spanish speakers (270 cases) are somewhat closer to those in the Mexican group (190), with the remaining groups far behind: Argentina (122 cases), Chile (110) and Cuba (102).
From these findings, it is also clear that the speakers from Spain also use most mitigating strategies per speech act.Thus, for these speakers, only 42% of speech acts contain a single mitigating strategy; by contrast, this is notably higher in the other groups: Argentina (55%), Mexico (62%), Chile (63%) and Cuba (74%).The speakers from both Spain and Argentina use two mitigating strategies in a speech act 30% of cases; by turn, this figure is considerably lower in the other groups: Mexico (20%), Chile (18%) and Cuba (15%).The remaining percentages for each dialect are of speech acts containing three or more mitigating strategies.Of note, only speakers from Spain use of seven strategies in one speech act, with lower maximums observed in the other groups: Mexico (6), Chile (5), Cuba (5) and Argentina (4).
In sum, these findings on the frequency of use of mitigators show a marked difference between the speakers of European and Latin American Spanish in the data.Of the latter areas, the Mexican speakers are seen to be most proximate to those of Spain, with the remaining varieties presenting lower levels.
Graph 2, below, illustrates the analysis of the mitigating function in speech acts as percentages and according to geographical area.The most notable differences in terms of the functions of mitigation are marked in Graph 2 with spotted bars.The first and second bars represent uses where mitigation affects only self-image or hetero image, respectively, and the third bar show cases where both faces are protected.The varieties are presented in ascending order according to the number of uses of mitigation with the function of self-image.As can be seen, speakers from Chile (28%) and Spain (29%) make the least use of mitigation to protect the selfimage.By contrast, speakers from both these countries coincide in making use of mitigation for reasons of politeness, that is, for the 'hetero image' and 'both images' categories, with the totals for the second and third columns for both varieties exceeding 70%.We must be noted that the 'both images' category includes 'hetero image' , which means that, being a polite use, it protects the speaker's own image at the same time as that of the interlocutor(s).See an example below in (7) (Section 4.2,below), repeated here for convenience: "Those dudes are gonna go bankrupt because of you (laughs)".The speaker uses laughter with a mitigating aim because she is threating her interlocutors' faces (hetero image).At the same time, she protects her face so as not to be considered a bad friend (self-image).
The results for each function in the Mexican and Cuban groups are similar.Speakers from both areas express an even distribution of these functions (each function is around 30-37%).By contrast, the Argentinian speakers show the greatest difference with regard to the other groups, especially compared to those from Chile and Spain.For the speakers from Argentina, 50% of mitigators serve self-image goals, whereas the polite uses (hetero-image and both images) represent 50% in total, a third less than in the other varieties.Thus, the speakers from Argentina in our data show the least allocentric and altruistic use of mitigation.
Turning to the use of mitigators according to gender, Graph 3, below, shows that when we take the whole dataset, some 500 minutes of conversation in Spanish, men use strategies of mitigation more (58%) than women (42%).

Men 58%
Women 42% This finding may seem surprising.According to popular belief, and also reported in some studies (García 2003, Molina 2005, Briz 2007), women are typically considered to be more cooperative in conversation and tend to be more polite, using more face-protecting strategies with regard to other interlocutors than men.Such a result has also been reported in other studies of geographical varieties of Spanish (studies based on interviews, not colloquial conversation): Las Palmas de Gran Canaria, Spain (Samper 2017), Monterrey, Mexico (Flores and Ramírez 2015) and Barranquilla, Colombia; (Torres 2020).
In light of our present findings, then, a change in this trend might well be taking place.
In future studies, a useful extension of the analysis would involve a greater number of sociolinguistic factors, towards gaining a greater understanding of the variables at play here.For example, the higher use of mitigation by men than women was also seen in previous studies based on corpora of interviews of the upper-class sociolect of Valencia, Spain (Albelda 2013), Madrid, Spain (Cestero 2017) and Santiago de Chile (González 2021).In these three cities it was also found that younger men (20-34 years) used strategies of mitigation Frequency of mitigation by gender in each variety the most, which is in line with our present study.Clearly, this issue merits further investigation.
If we break down our results by gender for each geographical area (Graph 4, above), we can see that, in speakers from almost all the areas, the percentage of mitigated acts is higher for men than for women.The differences are especially striking in the cases of Spain, Cuba and Mexico: in fact, the Cuban male speakers here mitigate twice as many speech acts as the Cuban women.This trend does not apply to the group from Chile, where mitigation is almost the same across the sexes; meanwhile, the Argentinian group once again stands out, in that women use mitigation more than men.
Finally, we turn to the functions that mitigation serves in each variety for men and women.From the group of graphs below, collected in Graph 5, it can be observed that in the results from several groups (Cuba, Argentina and Spain) women make far greater use of mitigation to protect their self-image than men (see examples 4 and 5 in Section 4.2, below).In these same three areas, men use far more mitigation than women to protect the face of others (hetero image, see example 10 in Section 4.2, below).By turn, in the groups from Mexico and Chile these proportions are inverted regarding the function of self-image (it is more frequent in men, though not strikingly so) and hetero image (more common in women).
It must be borne in mind that the frequencies here present certain limitations; our data pertain only to young adult speakers with a sociolect that can be supposed to reflect their high level of (university) education.Likewise, our data are drawn from conversations in just only one or two cities in each country.Further studies might usefully include data from colloquial conversations in other sociolectal variants, from a greater number of regions, and with more speakers from each area.

Type of Mitigation Strategies. Speaker's and Hearer's Face Exposure in Each Geographical Variety
In the 100 minutes of speech from each area, mitigation has been analysed according to the twenty types of strategies described in Section 3.2, above.In some cases, these strategies have been subdivided into variants.Table 2, below, sets out the raw findings for each variant according to geographical variety, divided into two parts.Below that, the results are presented in terms of descending frequency of the absolute frequencies for each variety (Figure 1).
Table 2 can be read vertically or horizontally.When read vertically some differences can be observed.For instance, the mitigating strategy of laughter (strategy 3) is far more frequent in the conversations from Chile than in any other variety, despite the fact that these speakers' total use of mitigators is lower than in the case of the Mexican and Argentinian speakers.Question tags and particles for controlling interaction 7 Laughing 13 Laughing 23 Laughing 8 Di fusers of meaning 21 Di fusers of meaning 11 Justi ers of the content 14 Justi ers of the content 13

Terms of address the unterlocutors 35
Figure 1 The six most frequent mitigating strategies in each geographical area In the Mexico group, justifiers of content (strategy 14) and objectivisation modals (strategy 20) are far more frequent than in the case of conversations from Spain, despite Spaniards having a higher total use of mitigators.In addition, the Cuban speakers use more terms of address than those from any of the other four varieties, although they also exhibit the lowest absolute use of mitigators.Horizontally, we can see the most frequent mitigating strategies in speakers from each group.In order to interpret this information, the six most common mitigating strategies from each geographical area have been extracted in Figure 1.
Diffusers of meaning and justifiers of content are both among the top six most common strategies in all the varieties, although with different weighting.More specifically, diffusers of meaning are the most commonly used strategy by the Argentinian speakers, and the second most frequent by those from Chile.The diffusor of meaning par excellence, used in speakers from all the dialects analysed, is como (que).Argentinians also used tipo, a particle that does not appear in our data for other dialects, although the Diccionario de americanismos also registers its use in Spain and Chile.In the Spanish speakers the forms en plan and rollo are found, but these do not appear in speakers from any of the other areas.Finally, in Figure 1, above, those strategies marked in red font are strategies that are not among the most commonly used by speakers of any other variety in the analysis.Of note here is the considerable difference in the Spanish conversations regarding ellipsis in the conclusion and in suspended structures (second most frequent), as well as the use of downgrading quantifiers (third most frequent).These structures have a very low incidence in all the other groups.
Graph 6, below, sets out the results in percentages of the general categories of mitigation, these being groupings of the 20 specific strategies analysed (see Section 3.2, above).As we have noted, these categories form a continuum, indicating the level of the speaker's commitment on the one hand, and on the other the exposure of the speaker's face to social and relational risks in the communicative interaction.The range of the categories goes from greater to lesser commitment on the part of the speaker, and consequently from greater to lesser face exposure.Likewise, from category 1 to category 7, the mitigating strategies used focus progressively less on the speaker's face and more on appeals to and the involvement of the addressee's face.8 Graph 6 illustrates the differences between speakers of each variety, and allow a more detailed view of the patterns of use of mitigation in these conversations from various locations in the Spanish-speaking world.The results can be read from left to right, with those varieties on the left being the ones in which mitigation focuses more on a greater exposure of the speaker's face and the action or inaction of the I-speaker (above all in Chilean speakers, and more moderately so for speakers from Argentina).At the centre of Graph 6 is the group from Spain, in that most of the mitigating strategies in these conversations fall within the two intermediate categories: category 6 (resources that involve the addressee in what is said or done) and to a lesser extent category 4 (minimisers and blurrers).Speakers from Spain thus represent the midpoint as regards face-exposure of speakers and hearers, whereas the Mexican and Cuban speakers (in contrast to those from the two southernmost countries) clearly opt for mitigating strategies which focus largely on the action or inaction of the addressee.
The Mexican speakers present the highest percentage of defocalising and impersonalising mitigating strategies, and thus this group is situated towards the right on the continuum, representing mitigating strategies that focus mainly on the action or inaction of the addressee and involving the greatest exposure of their face.
The Cuban group occupies the rightmost position in Graph 6, presenting as it does the greatest preference for mitigators centred on the hearer.It can be observed that 35% of the mitigating strategies in the Cuban data involve the hearer, which can be broken down into terms of address (27%), question tags (6%) and ellipsis of the conclusion (1.5%).Also, we note that defocalising and impersonalising strategies represent the second highest percentage of use in all five dialects (13%).

Graph 6
Results of general categories of mitigation 14 11.9 14 13.6 6.9 Graph 7 uses the same findings on the general mitigation categories seen in Graph 6, but illustrates differences between speakers from the geographical areas.These are organised according to which interlocutor is most affected in terms of face.
Graph 7 shows that in the continuum from more to less speaker commitment, the Chilean speakers (blue line) tend to use the greatest number of mitigating strategies on the left of the continuum (categories 1, 3, 4), whereas Cubans (black line) tend to use more strategies on the right (especially category 6, but also 4 and 5).The Argentinian group prefers mitigating strategies located in the middle of the continuum, favouring categories 3 and 4 (downgrading, minimising and blurring) that affect the content of the messages.Meanwhile, the groups from Spain and Mexico make greatest use of mitigators in the central categories and those on the right of the continuum; in both of these groups there is a relatively greater proportional distribution of mitigation categories overall, although with a trend towards the use of category 6 (more hearercentred) on the right.
The Graph 7 also shows that speakers from all five varieties make use of category 4 (resources that minimise or blur what is said), and hence we can claim that this category seems to be generalised as a means of expressing mitigation in our data from five points in the Spanish-speaking world, and perhaps might be generalisable here.
In conclusion, different trends in the use of mitigating categories can be seen depending on which interlocutors are more affected in terms of face.Such trends, although very general and requiring further study, can be summarized as follows.The Chilean speakers tend to take more risks in matters of the I-speaker's face, in that they use more mitigators from category 1.By contrast, and like the Argentinian speakers, they make almost no use of mitigators that affect the hearer's face (categories 6 and 7), and yet they also tend to use the more neutral categories in terms of communicative figures (categories 3 and 4).By turn, the speakers from Spain, Mexico and Cuba also show a relatively high use of the categories in the centre of the continuum, which focus on the content of the message.Also, these latter three groups all make notable use of category 6, in which the speakers expose their own face to very little risk, while exposing to the hearer's face to risk.

Concluding Remarks
The conclusions to be drawn from this study must be taken with a degree of caution.Despite the relatively large size of the sample, the data in fact cover only a small part of the sociolectal variety of each geographical area.In order to make broader generalisations as to the sociolectal patterns of mitigation in spontaneous, relaxed conversations, the sample would need to be extended and widened considerably, with more sociolectal groups (other age groups, different levels of education) and more dialects and subdialects of Spanish.
A further limitation of the present study arises, paradoxically, from one of its strengths.The study of a pragmatic phenomenon in free, spontaneous conversation enables us to perceive how it is used in its real context.In colloquial conversation, mitigation arises due to genuine conversational need.It occurs in a wide variety of situations and interpersonal relationships, among participants from different social and functional backgrounds and with differing levels of knowledge, and can serve an extensive range of communicative purposes.This wide variety of contexts, itself characteristic of colloquial conversation, is not something that can easily be controlled for in research (contrary to data from interviews based on questionnaires, for example).Thus, from the current findings we cannot be absolutely certain whether the observed sociolectal variation (between men and women), dialectal variation (speakers from five geographical areas) and linguistic variation (the mitigating strategies used) have their origins solely in the factors relating to these specific types of variation, or whether factors associated with situational and conversational variation might also be at play.
The use of a larger or smaller number of mitigators in these data might have been caused by a range of factors: references in the conversations to matters Downloaded from Brill.com 12/28/2023 09:15:18PM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/that are delicate or uncomfortable for some speakers, moments of difference of opinion or conflict, the greater presence of argumentative fragments, issues relating to age differences between speakers (here, 18 to 30 years), different socio-functional roles (for instance, host/guest), and many more.Future analysis of more speech samples will allow us to compensate for the lack of a wider variety of contexts in our present data; spontaneous speech data cannot be planned and controlled for experimentally in advance, but a greater number of contexts can indeed be achieved if larger amounts of data on spontaneous conversations is collected.This in turn will enable us to arrive at a clearer and more objective picture of sociolectal and dialectal variation.
So, from this initial study of the dialectal variation of mitigation in informal Spanish conversations, we can draw some tentative conclusions as regards our research questions, with the aforementioned caveats in mind.In terms of quantitative differences and similarities in the mitigation strategies used by speakers from the five varieties studied, the most striking finding is the greater use of mitigation in European Spanish compared to the four Latin American varieties analysed, with the Mexican speakers as the hinge point between the two continents as regards frequency of use here; our data suggest that Cuban and Chilean speakers are more moderate in their use of mitigation.
We have also found that, in overall terms, the men in our data use more mitigators than the women.This is most evident in the data on Cuban, Spanish and Mexican speakers.Meanwhile, the use is similar for Argentinian men and women, whereas Chileans tend slightly towards the inverse.
As regards qualitative differences in the use of mitigating strategies, we have found differing preferences in each variety.The speakers from Spain and Mexico exhibit a notable preference for question tags and particles as devices for controlling interactions, those from Argentina and Chile opt more frequently for diffusors of meaning, whereas the Cuban speakers prefer terms of address very significantly.Also, speakers of all the varieties share two of the most common strategies: diffusors of meaning and justifiers of content.
Finally, with regard to the exposure of the face of speakers and hearers arising from mitigation strategies, we found that the five varieties under examination span the entire continuum.Speakers from Chile and Argentina both show a greater tendency to use mitigators that guide the action or inaction away from the I-speaker, this also in the communicative content itself.In the conversations from Mexico and Cuba, on the other hand, the tendency is to use mitigators that direct the action or inaction towards the you-addressee.Speakers from Spain are situated somewhere between these two extremes, but with a greater tendency to use mitigators that involve the addressee.

Table 1
Geographical distribution and size of the conversation fragments analysed that repair or correct what is said or done 1. Apologies (I'm sorry to tell you that …) 2. Reformulations or corrections of what has previously been said (That was rude, Well …, That wasn't the best way to put it) 3. Prosodic or paralinguistic elements (often laugher) II.Resources that limit or restrict what is said or done 4. Constructions that limit an opinion to a particular person or to a certain field (In my opinion, To me) 5. Conditional constructions that restrict what is said (If you don't 19.Impersonalisation through reported speech (As my wife says, "pay no attention to those who don't deserve it") 20.Objectivisation using modal and evidential particles (The truth is that; indeed; as everybody knows; apparently, etc.) Downloaded from Brill.com 12/28/2023 09:15:18PM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/V. Resources that justify the speaker or the hearer 14.Justifiers of the content and of the act of saying, justifiers using I thought that … (Oh, I was late, there was a lot of traffic) VI.Resources that involve the addressee in what is said or done 15.Ellipsis in the conclusion or suspended structures (When I heard that rubbish I was like …) 16.Question tags and discourse particles for controlling the interaction (in Spanish, ¿no?, ¿verdad?, ¿sabes?, ¿viste?) 17.Terms of address and appellative formulas (in Spanish, mija, hombre, oye, wey, etc.) VII.Resources that impersonalise and defocalise 18. Impersonalisation by hiding the first person within the second person for a general interlocutor or when expressing a widely-held opinion (You can't just sit there without …)

Table 2
Mitigating strategies by geographical area