Characterization Frames Constructing Endoxa in Activists’ Discourse About the Public Controversy Surrounding Fashion Sustainability

This paper investigates the relationship between characterization frames and argumentation in activists’ discourse about the public controversy surrounding fashion sustainability. While previous studies proposing an argumentative approach to frames have acknowledged that frames are related to underlying implicit premises, how frames select certain implicit premises still needs to be systematically explained. Therefore, drawing on a theoretical framework combining Pragma dialectics (van Eemeren 2010) with the Argumentum Model of Topics an empirical analysis of a social media corpus has been performed in order to examine the connection between characterization frames and underlying implicit premises. This paper offers both a methodological and a theoretical contribution. From a methodological point of view, it offers linguistic-discursive tools for the analysis of characterization frames. From a theoretical perspective, it shows how characterization frames contribute to construct endoxa , i.e. underlying implicit premises, through argumentative patterns.


Introduction
This paper examines how social media activists use characterization frames to construct endoxa, i.e. shared cultural premises with an argumentative function, by taking into analysis the case of the public controversy surrounding fashion sustainability.
In recent years, the topic of fashion sustainability has become increasingly debated in the public sphere, both in terms of the environmental impact of the fashion industry and of the social concerns related to the condition of garment workers. The discussion around fashion sustainability is closely related to debate over climate change (see Carbonaro and Goldsmith 2015;Wallinger 2015), as the fashion industry generates emissions, pollution and waste in every phase of the production process of garments (Hibbert 2018). Moreover, the success of fast fashion has encouraged consumers to purchase a huge number of garments, which are then discarded after only a few uses, thus creating impressive amounts of textile waste (Niinimäki et al. 2020). In addition to these environmental factors, the exploitation of workers employed in the production of fashion, who often are denied fair wages and lack basic human rights (Hibberd 2018;Prentice et al. 2018), constitutes the other area of discussion.
According to Greco and De Cock (2021), this debate represents a case of public controversy, that sees the involvement of different actors such as public institutions, fashion companies, citizens, NGOs and small businesses. In line with Greco and De Cock (2021), I consider this public controversy as an instance of argumentative polylogue, i.e. a discussion which involves multiple players, positions and places (Aakhus and Lewiński 2017). While in this paper I focus on the argumentation advanced by one specific group of players of the controversy, i.e. fashion activists posting on social media during Fashion Revolution Week, their discourse is part of the overall polylogue. In fact, activists' discourse on social media is not a monologic action, as it arises in response to certain events which are related to the practices of other players in the controversy, for example to the Rana Plaza collapse in 2013.
In line with Greco and De Cock (2021) the multiple players involved in the polylogue hold different views in respect to sustainability at various levels, which often turn into argumentative misalignments, that is, incongruities in 1 3 the material starting points at the opening stage (van Eemeren and Grootendorst 2004) of the argumentative discussion. This paper starts from the assumption that one type of argumentative misalignment present in the controversy is to be found at the level of conflicting characterization frames (Shmueli 2008) held by the parties involved, i.e. the conflicting ways in which the conflicting players express their views of themselves and others; in particular, characterization frames refer to how parties interpret their own behaviour, that of others, as well as the relationship between others (Shmueli 2008(Shmueli p. 2051. Following scholars in conflict resolution studies (see Shmueli et al. 2006;Shmueli 2008), an analysis of the conflicting frames held by the different parties is crucial to understand the clashing points present in the controversy at issue. Therefore, identifying the conflicting characterization frames and determining the reasons underlying them is important to understand where the controversy sparks from. However, these studies do not clarify how characterization frames can be reconstructed, starting from discourse.
In this respect, argumentation can help to investigate the reasons underlying the characterization frames employed in the controversy. Building on studies proposing an argumentative approach to frames (Greco Morasso 2012;Bigi and Greco Morasso 2012), this paper aims at explaining the relationship between a specific kind of frame, i.e. characterization frames, as used by social media activists, and their underlying argumentative premises. In particular, this research focuses on clarifying which role characterization frames play at the level of argumentative premises, where by this term I mean material starting points and particularly endoxa, that are "general premises that are accepted by the relevant public in a specific argumentative situation" (Rigotti andGreco 2019, p. 2014).
Overall, this paper offers both a theoretical and a methodological contribution. From a theoretical perspective, it shows how characterization frames select certain material starting points (endoxa) through argumentative patterns, a pragma-dialectical notion that refers to a series of argumentative moves that occur in relation to strategic manoeuvring in a specific communicative activity type (van Eemeren 2016); in this case, within the discourse of social media activists. At the methodological level, this paper develops a refined definition of the concept of characterization frames to explain how these frames are linguistically activated in discourse.
Furthermore, this research adds to the growing body of academic works which consider social media to be argumentative (e.g. Goodwin 2020; Elliott-Maksymowicz et al. 2021). More specifically, as this study considers activists' discourse on social media as a form of political expression (Velasquez and LaRose 2015), it contributes to the existing literature about political argumentation on social media (see Lewiński and Mohammed 2015;Mohammed 2019).
The paper will develop as follows. In Sect. 2, I present the literature review about the argumentative approaches to frames and the notion of characterization frames as explained in conflict resolution studies. In Sect. 3, I explain the criteria for the corpus collection and for the procedure of analysis, which is composed of two different stages. Then, in Sect. 4 I discuss my findings, while Sect. 5 concludes the paper and outlines proposals for future research.

Argumentative Approaches to Frames and Framing
The notions of frame and framing have a rich history in academic literature across various fields, including sociology and linguistics (see Tannen 1979). From the perspective of argumentation studies under the pragma-dialectical approach, frames play an argumentative function. Following the classification of the studies on framing proposed by Dewulf et al. (2009), I distinguish between a first cohort of argumentation scholars (van Eemeren 2010; Fairclough and Mădroane 2020) that draw on an interactional understanding of the concept, usually adopting the term framing, and a second cohort (Greco Morasso 2012;Bigi and Greco Morasso 2012) that follow a cognitive conceptualization of frames, and who prefer to use the term frames. Argumentation scholars belonging to the first cohort see framing as an interactional co-construction (Dewulf et al. 2009) and quote definitions elaborated by authors including Goffman (1974) and Entman (1993). According to van Eemeren (2010), framing is employed in the strategic manoeuvring process, in which the protagonist tries to persuade the antagonist of the acceptability of a standpoint by balancing rhetorical effectiveness and dialectical reasonableness. Framing is linked to the aspect of presentational devices in strategic manoeuvring, as it aims at presenting facts and issues in a perspective that is not neutral, but that backs the standpoint advanced by the protagonist. In this view, framing draws attention to particular details of the reality at issue, while simultaneously shifting attention away from other details, thus helping to convey a certain interpretation of reality (van Eemeren 2010).
Also Fairclough and Mădroane (2020) use the term framing and focus particularly on the framing process. Drawing on the characteristics of framing outlined by Entman (1993), namely selection and salience, the scholars identify the framing process as a type of deliberative and practical reasoning, which has the end goal of orienting the audience 1 3 towards a certain course of action. According to Fairclough and Mădroane (2020), in the framing process the reasons deemed most salient for a certain goal are selected. Thus, framing is applied to highlight certain premises of the deliberation scheme, while overshadowing others, in order to guide the audience to a decision about what to believe or do. For example, if an action is framed as a "criminal and corrupt activity", the proposal entailing that crime frame will necessarily be rejected: framing generates "a set of inferences, which are transferred to the understanding of a new situation, leading to particular conclusions" (Fairclough and Mădroane 2020 p. 135).
The second cohort of argumentation scholars I consider here view frames as cognitive representations (Dewulf et al. 2009) and adopt the linguistic definition of frames developed by Fillmore (1976). In this perspective, frames constitute the structures that individuals use to understand the reality surrounding them and that are activated though certain specific frame-activating words. For example, the word "buy" activates a commercial frame, a scenario in which specific elements come to life to the mind of the hearer, including roles like buyer and seller which are evoked by simply activating the frame without mentioning them (Fillmore 1976).
Through a fine-grained analysis combining pragmadialectical notion of strategic manoeuvring (van Eemeren 2010) with the Argumentum Model of Topics (see Rigotti and Greco 2019) in news discourse, Greco Morasso (2012) claims that frames are connected to material starting points in argumentation and especially to cultural premises, called endoxa in the Argumentum Model of Topics' reinterpretation of the Aristotelian concept. According to Greco Morasso (2012, p. 200), frames can be activated as material starting points for arguments: the choice of certain frames "allows the selection of adequate material starting points, that is, shared views and values, which can be used to construct arguments". For example, in news discourse, choosing to report the death of a person as an incident or as the result of a human action select certain different material starting points, i.e. different endoxa, that emerge clearly when performing the reconstruction of the argument (Greco Morasso 2012).
On the relationship between frames and implicit premises in argumentation, i.e. material starting points, Bigi and Greco Morasso (2012) suggest that the notion of semantic frames can help to understand the connection between the level of implicit premises and the lexical level of argumentation. In fact, given a frame that conveys a specific interpretation of a determined situation, in argumentative discussions it is possible to derive from that selected frame a number of endoxa that concern what should be believed or what should be done in respect to that determined situation (Bigi and Greco Morasso 2012).
While all the contributions outlined in this section have recognized that the choice of selecting a specific frame over others is part of an overall procedure aimed at persuading the audience of accepting a certain standpoint, only the authors belonging to the second cohort have investigated the role of frames in relation to underlying implicit premises. More precisely, Greco Morasso (2012) and Bigi and Greco Morasso (2012) have claimed that frames select, or activate, material starting points in argumentation, specifically endoxa.
However, the precise argumentative dynamics that show how frames activate endoxa still need to be systematically explained. In this paper, building on this second group of contributions, I advance the reflection on the relationship between frames and underlying implicit premises, with a focus on a specific kind of frames, i.e. characterization frames, that originally comes from conflict resolution studies. Thus, in line with Greco Morasso (2012) and Greco Morasso and Bigi (2012), I adopt a view of frames as cognitive concepts in the Fillmorean sense. Moreover, my research combines pragma-dialectics with the Argumentum Model of Topics, as I consider this integration important for grasping the dynamics underlying the relationship between characterization frames and implicit premises.

The Concept of Characterization Frames in Conflict Resolution Studies
In situations of controversy like the one examined in this paper, the notion of frame acquires great importance. According to authors in conflict resolution studies (Shmueli et al. 2006;Shmueli 2008), reconstructing the conflicting frames present in a controversy can help to understand how conflict emerges. In this paper, I consider a specific type of conflicting frames, i.e. characterization frames (see Elliott 2003;Kaufman and Smith 1999;Shmueli et al. 2006;Shmueli, 2008), since I believe that the analysis of this type of frames is crucial to capture some essential points in the controversy, related to both the important values held by the parties and to the view they have of other players. Following Shmueli et al. (2006), characterization frames concern the attribution of positive or negative features to the parties involved in the conflict. More precisely, for Kaufman and Smith (1999, p. 171) such characterization refers to the players giving an evaluation of others' behaviour, attitudes, motivation and trustworthiness. In this paper, I follow Shmueli's definition (2008, p. 2051) which is centered around one specific element, namely behaviour: "Characterization frames reveal how parties perceive their own behavior, characterize the behavior of others, and assess relationships among the parties". When players talk about their own features and about how they see their ability of intervening in the conflict, they activate self-characterization frames (Kaufman and Smith 1999).
Closely related to the category of characterization frames are identity and value frames (Shmueli 2008), which describe how the parties view themselves in the conflict situation, as well as how they perceive fundamental values. According to Shmueli (2008), the parties' sense of identity results from the intersection between their self-images, their interests and their belonging to certain groups. Moreover, the conflicting parties' identity frames tend to become more intense as the dispute escalates (Elliott 2003).
From the definitions of characterization and identity and values frames, these two categories appear deeply connected, as acknowledged by Shmueli et al. (2006). In fact, when a certain group characterizes the other party, it is also implying a definition of its identity; at the same time, when some players describe how they identify themselves, they are simultaneously offering a self-characterization. For this reason, I propose to adopt here an understanding of the characterization frames category which partially overlaps with identity and values frames.
While I acknowledge that characterization frames are originally related to conflict resolution studies, for my research aim I focus on the concept of characterization frames, leaving to future research a deeper understanding of the specific role they play when they become conflicting characterization frames, i.e. when considering the conflicting positions of the different players (see Sect. 5).
Although the contributions outlined in this section introduce the concept of characterization frames and offer some examples of what may constitute a characterization frame, they lack the fine-grained linguistic tools necessary to systematically analyze them in a corpus. In fact, these studies do not explain how characterization frames are represented in discourse, which makes it difficult to identify how they are linguistically activated in a corpus.

Data Collection
In order to empirically accomplish my research aim, I collected a corpus composed of 400 social media posts, 200 Instagram posts and 200 Tweets, all containing #fashionrevolution. The posts were published in English during Fashion Revolution Week 2020, which took place between 20th and 26th April.
Fashion Revolution (https:// www. fashi onrev oluti on. org/) is an activist organization that calls for transformation of the fashion system into a more environmentally and socially fair and social industry. This organization was created following the Rana Plaza disaster in 2013, when a building in which garments for global brands were produced collapsed, killing over a thousand workers (Sádaba et al. 2019). Every year, in correspondence of the anniversary of the Rana Plaza, the movement organizes Fashion Revolution Week, an annual campaign which encourages people to participate in the transformation of the fashion industry.
On social media, participation to the campaign happens through the posting of different hashtags like #fashionrevolution, which constitutes an instance of "rallying hashtag calling for collective action" (Karamalak and Cantoni 2021, p. 243). Drawing on this definition, I consider all participants to the #fashionrevolution 2020 campaign as social media activists. This cohort represents a rather heterogeneous group, as it includes not only members of the Fashion Revolution organization, but also small brands and citizens (see Greco 2022).
Overall, social media activists have mainly two objectives they intend to achieve with this campaign. First, they aim to create awareness about the issues related to the fashion industry and promote change in behaviour given that "Currently, there is a lack of understanding and appreciation of the true cost of clothing" (Fashion Revolution n.d.). Then, they want to push public institutions to take action: on its website, Fashion Revolution states that "we are calling for new laws that require businesses to conduct due diligence on living wages" (Fashion Revolution n.d.). From an argumentative perspective, following Mohammed (2016), these aims can be seen as instances of the intrinsic function of the act of arguing, that is, the rhetorical goal of argumentation, which is to persuade another party. As it is often the case in public political arguments, in the context examined activists do not want to persuade who might appear their primary interlocutor, i.e. big brands, but a third party, namely the general public and legislation bodies.
The corpus was obtained through a hashtag-based extraction performed by two providers, who operate in accordance with the rules of the social media platforms: the Centre de Traitement Automatique de Langage of UCLouvain for Tweets (see Naets 2018) and Picodash (www. picod ash. com) for the Instagram posts. For the research aim of this paper, I consider only the verbal component of the posts, namely the Tweets and the captions of the Instagram posts, leaving out the multimodal dimension.

A Refined Definition for the Analysis of Characterization Frames in Discourse
To understand how characterization frames are related to material starting points in argumentation, it is first needed to clarify how characterization frames are activated in discourse. To this end, I propose a refinement of the definition given by Shmueli (2008), based on the intuition that frames are activated through certain frameactivating words (Fillmore 1976). According to Shmueli (2008), what gets framed in the case of identity and characterization frames are, broadly speaking, descriptions and attributes of the players involved in the controversy. Drawing on this consideration, I proceeded both inductively and analytically by annotating in my corpus all the linguistic expressions that were either referring to the author of the text (in this circumstance, they are instances of self-characterization) or to the six main groups of players involved in the polylogue, i.e. garment workers, small businesses, big brands, activists/NGOs, public institutions and consumers. These groups of players have been identified starting from Greco and De Cock (2021), with the addition of two groups which have not been recognized explicitly in their contribution: first, the garment workers, who constitute the core of the "social" side of the sustainability issue in the fashion industry, and whose voice in this paper I consider represented by the members of the Fashion Revolution campaign. The other group is the "activists" category, that Greco and De Cock (2021) do not explicitly call by this name, as they talk about NGOs; however, the activists' group has been recognized as one of the key players in the controversy in subsequent research (see Greco 2022).
From this inductive annotation, I developed a refined definition of the concept (see Fig. 1), according to which characterization frames refer to how the players in the polylogue describe themselves and others in three different ways: in terms of persona, stable behaviour and state/condition.
By persona, I mean the social character defining the identity of an individual or group. To identify how characterization frames associated to a certain persona are activated in discourse, I select portions of social media posts containing a definition of the identity of that group or individual. This is usually expressed through nouns and adjectives. For example, garment workers are characterized with terms such as "employees" or "countless anonymous members of the supply chain".
By stable behaviour, I refer to instances of characterizing actions, i.e. actions that are described as constituting part of the identity of the player due to their frequency or intensity. Thus, for this kind of characterization frames, I select parts of social media posts containing mentions of actions with qualities of frequency or intensity performed by players in the polylogue. Therefore, "Artisans using vintage silks, chiffons and organic cottons" would be classified as a characterizing action because of its frequency, whereas "Fast fashion exploits women" would be considered a characterizing action for the intensity of this act. The linguistic representation of characterization frames associated to stable behaviour usually contains verbs in active tense, often in combinations with adjectives and nouns.
By state/condition, I mean the resulting situation in which a certain player finds herself as the result of the stable behaviour of another player. Characterization frames associated to a state/condition are selected following the same reasoning explained for stable behaviour, with the difference that in the case of state/condition the player is the recipient, or victim, of the behavior of others. Linguistically, this type of characterization frames is often represented with verbs in the passive tense, combined with adjectives and nouns. For example, the expression "Workers forced to work in dangerous conditions" constitutes an instance of the discursive activation of characterization frames in the sense of a state/ condition in which garment workers find themselves as a result of the stable behaviour of others. The examples presented for each type of discursive activation show that characterization frames can be linguistically activated in multiple ways, through the use of nouns, adjectives, verbs in the active and passive tense, as well as through combination of all these elements. This makes it difficult to propose a clear linguistic definition of characterization frames, as they do not correspond to a specific lexical class and are not necessarily conveyed by a single term, especially when they are associated to a stable behaviour or a state/condition; on the contrary, in many cases they are activated through a combination of words or even through a whole sentence, such as in the expression "They gather and hand-stitch soft cotton with recycled saris and threads". In this sense, with reference to Fillmore (1976), it appears more adequate to talk about frame-activating expressions rather than frame-activating words.
A further consideration to be made is that the characterization frames of the different players are often intertwined, as the stable behaviour of one player, such as a small brand, can result in a certain condition lived by another, for example the good treatment of garment workers: this matches the original definition by Shmueli (2008Shmueli ( , p. 2051, who observed that characterization frames also reveal how parties perceive relationships among them. What this means is that the same discursive expression can contain a characterization frame of player A and at the same time a characterization frame of player B. For instance, the sentence "Every day we honor of our weaving artisans" would be entirely annotated as an instance of characterization frame of 'small businesses' (of the stable behaviour type), while the terms "weaving artisans" would be categorized as characterization frame of 'garment workers' (of the persona type).

Creation of Categories of Characterization Frames
Following the elaboration of this refined definition, I aggregated the single occurrences of frame-activating expressions for each group of players in categories of characterization frames, based on the predicate that I identified as common between certain frame-activating expressions. Here, drawing on linguistic semantics, I use the notion of predicate as explained in the Congruity Theory by Rigotti and Rocci (2001), according to which "A predicate is conceived ontologically as a possible mode of being, a general notion that subsumes more specific ontological distinctions such as those between properties and relations, states and events, actions and non-actions" (Rigotti 2005, p. 78).
For example, I identified expressions like "the ones who are suffering today because of fast fashion" and "underpaid and unpaid factory workers forced to work in dangerous conditions" as both containing the predicate, i.e. the mode of being, "to be exploited/suffering/abused" and therefore I included them in an encompassing category named "Workers being exploited/suffering/abused". Or again, both the frame-activating expressions "In this factory we take floor cuttings, mix them with water and sunlight, and create paper, stickers and our notebooks" and "by weaving everything by hand, with all-natural materials, in a small studio off-grid using solar energy" present a common predicate, that is, "undertaking virtuous practices" and thus have been included in the category "brands undertaking virtuous practices".
Thus, these categories are named according to the predicate that is common among a certain group of frame-activating expressions. For this reason, I claim that these categories of characterization frames are based on the recognition of overarching predicates, that are, general modes of being that summarize groups of frame-activating expressions.

Argumentative Analysis to Reconstruct Patterns
In the second phase of the methodology, I performed an argumentative analysis in order to explore which argumentative patterns were related to the different characterization frames, to then investigate which endoxa were constructed. In fact, I consider the notion of argumentative patterns as essential to reconstruct the argumentative dynamics through which characterization frames activate endoxa. To this end, I combined the pragma-dialectical approach (van Eemeren and Grootendorst 2004; van Eemeren 2010) and the Argumentum Model of Topics (henceforward: AMT), elaborated by Rigotti and Greco (2019).
In Pragma-dialectics, the concept of argumentative patterns is related to strategic manoeuvring and occurs within a certain communicative activity type which belongs to a specific communicative domain. Starting from the consideration that communicative activity types can be characterized argumentatively, van Eemeren (2016, p. 13) notices that, within a given communicative activity type, certain kinds of argument schemes will play a prominent role for achieving the desired goal. Then, he defines argumentative patterns as "A constellation of argumentative moves in which, in order to deal with a particular kind of difference of opinion, in defence of a particular type of standpoint a particular argument scheme or combination of argument schemes is used in a particular kind of argumentation structure" (van Eemeren 2016, p. 14).
For the purpose of this paper, I consider argumentative patterns as a useful tool for understanding which role characterization frames play in relation to argumentation in a specific communicative activity type, that is, social media posts by fashion activists, within the political domain. In this phase of the analysis, I considered only the most frequent categories of characterization frames, i.e. which contained at least three occurrences of frame-activating expressions. For each identified category of characterization frames, I analyzed the social media posts containing the frame-activating expressions belonging to that category, by completing an analytic overview (van Eemeren and Grootendorst 2004) of the standpoint and supporting arguments. This reconstruction permitted to identify where the characterization frame was located within the argumentation structure.
Then, I searched for the presence of argumentative patterns within each category of characterization frames. In my use of the concept of argumentation patterns for this paper, this meant I focused on determining the arguments schemes, or loci, according to the taxonomy proposed by Rigotti and Greco (2019) and the types of standpoint, i.e. evaluative, descriptive or prescriptive (van Eemeren 2010) which appeared most frequently for each category of characterization frames. For the concept of locus, I follow Rigotti and Greco (2019Greco ( , p. 2010, that define it as "the source from which arguments are taken" and the "ontological relation on which a given argument is based". All this detailed procedure of analysis was important to understand the connection of characterization frames to implicit premises and in general the relationship to argumentation. The following example illustrates this procedure, applied to the analysis of a Tweet: It's time to dump fast fashion! Did you know it takes 2700 litres of water to make just 1 T-shirt. Consider repurposing your garments and help the environment! […] #fashionrevolution In this Tweet, the frame-activating expression identified is 'fast fashion', which is related to the group of players 'Big brands'. This expression is included in the corresponding overarching category 'Fast fashion'. While in this case the frame-activating expression coincides with the overall category, in most cases the frame-activating expressions do not correspond to the overall category, even though they share the same predicate (see 3.2.2). The reconstruction of the argumentative structure for the part of tweet that includes the selected characterization frame is presented below. The number 1 indicates the standpoint, whereas 1.1. refers to the supporting argument (see van Eemeren 2010).

It's time to dump fast fashion 1.it takes 2,700 litres of water to make just T-shirt
In this example, the characterization frame is contained in the prescriptive standpoint "It's time to dump fast fashion!". The identified locus linking the standpoint to its supporting argument is a locus from termination-setting up (Rigotti and Greco 2019, p. 263), where the present situation in which "it takes 2700 litres of water to make just 1 T-shirt" is compared with a future possible world, that is, a world where less T-shirts are produced. Following the result of this comparison, the speaker advocates for terminating the current situation, inviting the audience to "dump fast fashion". This same procedure illustrated for this example has been applied to all instances of characterization frames found in the corpus.

Findings: Argumentative Patterns Underlying Characterization Frames
Before presenting the findings in response to the theoretical aim of this paper, i.e. understanding how characterization frames are related to endoxa through the analysis of argumentative patterns, in Table 1 I report the occurrences of frame-activating expressions that I found for each group of players mentioned in the corpus. Table 1 shows that characterization frames referring to garment workers and small businesses are much more frequent than those referring to other players. This may be because, in the Fashion Revolution campaign, activists are mainly concerned about (a) raising awareness about the conditions of garment workers and (b) presenting businesses they consider sustainable as alternatives to the current fashion brands. This last point is also in line with the consideration that the participants to the campaign do not exclusively consist of members of the Fashion Revolution organization, but also include small brands (see 3.1), who use self-characterization frames to position themselves in opposition to big brands. Interestingly, characterization frames related to public institutions appear only three times in the whole corpus. A possible explanation for this result is that, according to participants to Fashion Revolution Week, the characterization of public institutions is not a key aspect of the discussion; in other words, it is not a point about which the different players present incongruities.
Then, in Table 2I summarize the two main argumentative patterns that have been retrieved in association to characterization frames categories and that I will discuss in the following sections. Table 2 includes the group of players, the characterization frames categories and the related argumentative patterns, i.e. the standpoint, the type of standpoint, the locus, the argument and the location of the characterization frame for each category. From Table 2, it emerges that the two most recurrent argumentative patterns are the following. First, the one containing the evaluative/descriptive standpoint "We/this business can legitimately say to be part of Fashion Revolution" and the locus from definition, present in 'garment workers' and 'small businesses'; second, the one composed of the prescriptive standpoint "You/we should undertake a certain action towards fast fashion/big brands" and the locus from termination and setting up, associated to 'big brands' and also partially 'garment workers', in which case the standpoint is "A revolution is needed because currently workers are being exploited/suffering/abused".
Other identified argumentative patterns for the different groups of players are reported in the Appendix, in which I include the less frequent argumentative patterns that were in found in relation to at least two categories of characterization frames. In the Sect. 4.1 and 4.2, I will discuss how the characterization frames contained in the two most recurrent argumentative patterns are related to endoxa.

Characterization Frames Constructing Endoxa Through Argumentative Patterns Containing the Locus from Definition
In this section, I show how the characterization frames contained in the most recurrent argumentative pattern shown in Table 2, which is associated to 'garment workers' and to 'small businesses', is linked to the construction of endoxa. This argumentative pattern includes the following elements: a locus from definition and an evaluative or descriptive standpoint "We/this business can legitimately claim to be part of Fashion Revolution", usually implicit, supported by an argument containing a characterization frame. The argument varies slightly between the two groups of players: for 'garment workers', the argument is of the kind "I/ we/this business involves [workers treated in a certain way, e.g. being treated fairly", whereas for 'small businesses' it is of the kind "I/this business [is sustainable]/operates sustainably. The following Instagram post offers an example of this pattern.

Fashion Revolution Week is coming to a close but let us continue to ask, 'Who made my clothes?' Scroll through to see some of the empowered women we partner with to make our collections-@khaloom handloom weavers and @_saheliwomen garment workers-who are provided fair trade wages, safe and healthy work hours and conditions, and social safety nets like health care. […] #fashionrevolution.
In this Instagram post, of which I report here a shortened version that includes only the relevant paragraph for the analysis, I highlighted the frame-activating expression "garment workers-who are provided fair trade wages, safe and healthy work hours and conditions, and social safety nets like health care" (of the state/condition type) that I included in the overarching category 'Workers being treated fairly', under the group of players 'garment workers'. The resulting reconstruction is presented below, with the implicit standpoint, which is unexpressed in the post, indicated in parentheses.
(1) (We can legitimately claim to be part of Fashion Revolution)

We partner with empowered women to make our collections-@khaloom handloom weavers and @_ saheliwomen garment workers-who are provided fair trade wages, safe and healthy work hours and conditions, and social safety nets like health care.
The implicit standpoint "We can legitimately claim to be part of Fashion Revolution" can be reconstructed thanks to the knowledge of the context, that is, the Fashion Revolution Week, which has the aim of raising awareness and promoting change (see 3.1). In this context, fashion brands are indicated as responsible for the current negative situation, as stated on the Fashion Revolution website, which reads "Brands are avoiding the realities of climate breakdown by continuing to pursue extractive business models and greenwashing their way to sustainability (Fashion Revolution n.d.). The participants to the campaign, who often are small businesses or individuals concerned with sustainability, want to clearly differentiate themselves from these brands, those that I called "big brands" in this paper. Therefore, they present their practices and views to justify their claim of legitimate belonging to the radical transformation of the fashion industry advocated by Fashion Revolution.
After having identified standpoint and argument, I applied the AMT for reconstructing the internal functioning of the argument, called by Rigotti and Greco (2019) inferential configuration. Thanks to the AMT, both the procedural premises of the argument, i.e. locus and maxim, and the material ones, i.e. endoxon and datum, are made explicit. For the procedural component, I adopt here the same maxim applied for definitional loci by Schär (2017, p. 185), while I derive the material component from the text of the examined Instagram post. Taking into consideration the overarching category to which the characterization frame present in this example belongs, I reconstructed the inferential configuration of the argument containing "workers being treated fairly" in Fig. 2.
This inferential configuration shows that the characterization frame "Workers being treated fairly" is contained in the datum; and that the endoxon, i.e. the shared cultural premise, is the following: to belong to the species "Being part of Fashion Revolution", a business or an entity needs to possess the essential property "Treating workers fairly". In For each group of players, the categories are reported according to their frequency in the corpus, from the most frequent to the least frequent 1 3 other words, the endoxon reveals that "Being part of Fashion Revolution" necessarily means "Treating workers fairly". The inferential configuration reconstructed for this specific category of characterization frames can be generalized to all the categories in the 'garment workers' group that, as indicated in Tables 2, present the same argumentative pattern. This step can be performed simply by proposing an encompassing formulation of endoxon and datum, as shown in Fig. 3.
Thus, the general endoxon for this argumentative pattern reveals that for an entity "Being part of Fashion Revolution" means possessing a certain "characterization frame x of workers". Drawing from this, all the categories of characterization frames presenting this pattern included in Table 2 contribute to constructing this endoxon, i.e. what it means to be part of Fashion Revolution: • To work with artisans; • To have workers being treated fairly; • To be supporting workers in a vulnerable condition; • To be undertaking practices supportive of workers; • To be involved with workers with positive characteristics; All these categories form part of the shared cultural premise about the meaning of "being part of Fashion Revolution", which in other words means to take part in the radical transformation of the fashion industry.
The same line of reasoning can be applied to the group of players 'small businesses'. Also in this case, the categories of characterization frames construct an endoxon that is about establishing what it means to "Being part of Fashion Revolution. It results that to take part of Fashion Revolution for a small business means: • To be ethical. • To be sustainable. • To be a business with a focus on the environment, • To be local or small. • To undertake virtuous practices, in particular treating workers fairly and using sustainable materials; • To support workers; • To work with ethical partners; • To be proud of their sustainable practices; • To have a humanitarian goal; • To be revolutionary; • To be transparent.
To summarize, the endoxon that characterization frames construct through the illustrated argumentative pattern is that a player can legitimately claim to be revolutionary if they possess these mentioned essential attributes and characteristics. By using these characterization frames to refer to a certain view of garment workers and to the qualities of small businesses, activists are pointing out that the big brands do not possess these attributes, thus they cannot claim to be participating in the radical transformation of the fashion industry; in other words, that they cannot claim to be sustainable.

Characterization Frames Constructing Endoxa Through Argumentative Patterns Containing Locus from Termination and Setting up
The second recurrent argumentative pattern that I discuss in detail is related to both 'big brands' and 'garment workers'. In the case of 'big brands', it consists of the prescriptive standpoint "You/we should undertake a certain action towards fast fashion/big brands", supported by an argument reporting a negative situation that involves big brands' practices, with a locus from termination/setting up. In this pattern, the characterization frame is found in the argument for certain categories and in the standpoint for others; thus, its location varies. On the contrary, for the categories associated to 'garment workers' the characterization frame always appears in the argument. For this group of players, the argumentative pattern presents the same locus, but seems different because it contains the evaluative standpoint "A revolution is needed"; however, in this section I will argue that it is just a variation of the pattern found in relation to 'big brands'.
To illustrate this argumentative pattern as it appears in the categories related to 'big brands', I present the analysis of the Instagram post below, of which I report only the parts relevant to this section.
[…] We love fashion isn't it? But we don't want our clothes to exploit people or destroy the planet. Do you Fig. 2 Inferential configuration of the "Workers being treated fairly" category know that the fashion/textile industry is one of the major polluting industries in the world? Fast Fashion Exploits Women. As feminist we can't accept that. The majority of people who make our clothes are women. These fearless women risk their lives everyday to feed their families. They work long hours and endure hazardous fumes in hardly-upto-code buildings. Sexual harassment and abuse are also very common during work inside the factories.
The two frame-activating expressions highlighted in the Instagram posts refer to the group of players 'big brands'. Following the methodology explained in Sect. 3.2.2, I included the linguistic expression "Fast fashion exploits women" (of the stable behaviour type) in the overarching category "Big brands exploiting/hurting workers" and "Fast fashion" (of the persona type) in the category "Fast Fashion", as in this case the predicate coincides with both the discursive expression of the characterization frame and the overarching category. The following reconstruction illustrates the argumentative structure of this example.
(1) You should Finally tell Fast Fashion that you Deserve Better

Fast Fashion Exploits Women
From the reconstructed argumentative structure, it can be noticed that the frame-activating expressions appear both in the standpoint and in the argument, as it was previously mentioned. Moreover, in this example the standpoint is explicit and invites the reader to take action towards fast fashion and specifically to give up the purchase and consumption of this kind of brands. This prescriptive standpoint is supported by an argument that depicts a negative situation, that is, the exploitation of women. Thus, standpoint and argument are linked by a locus from termination and setting up: from the evaluation of the current state of affairs, it derives the invitation to terminate the present negative situation. In Fig. 4, I reconstruct the inferential configuration of the argument containing "Big brands exploiting/hurting workers", i.e. of the overarching category to which the characterization frame "Fast fashion exploits women" belongs.
The AMT inferential configuration shows that the endoxon contains a negative evaluation of the stable behaviour through which 'big brands' are characterized; in this case, the exploitation and hurting of workers. The same endoxon is constructed for the other category of 'big brands' that presents the same argumentative pattern with the characterization frame in the argument, i.e. "Fast fashion hurting the environment", and also for the categories in which the characterization frame appears in the standpoint, i.e. "Fast fashion" and "Big/global brands". In these cases, the characterization frames appear in terms of persona and provide a vaguer characterization of the practices of big brands.
The negative evaluation of big brands' practices present in the endoxon is consistent with the overall aim of the Fashion Revolution campaign, which is to invite all members of the public to contribute to the radical transformation of the fashion industry. According to activists, this revolution will happen by taking action against the actors deemed responsible for the present negative situation, that is, the big brands. In other words, the negative evaluation of big brands' practices becomes the cultural shared premise which justifies the invitation to take action against big brands. Figure 5 shows the general inferential configuration of the argumentative pattern containing the locus from termination and setting up for all categories in 'big brands', in which the characterization frame is found in the argument.
For the cases in which the characterization frame is located in the standpoint, the inferential configuration would be almost identical: the datum would be "Certain practices undertaken by big brands are in place", the endoxon "Practices undertaken by big brands are negative" and the final conclusion, i.e. the standpoint, would be "You should undertake a certain action towards characterization x of big brands".
The argumentative pattern that has been discussed is also found in a slightly different form in the 'garment workers' group, as stated at the beginning of this section. In fact, the evaluative standpoint "A revolution is needed" is closely related to the prescriptive standpoint "We should undertake a certain action towards fast fashion/big brands". Given the context of the Fashion Revolution Week, I argue that this evaluative standpoint could be read as well as a prescriptive one, because from saying that a revolution is needed, it can easily be derived that we should take action to achieve the advocated revolution (see Soria-Ruiz et al. 2022). Moreover, this pattern includes the argument containing the characterization frame "Workers being exploited/suffering/abused", which is the same analysed in Fig. 4, formulated from the perspective of garment workers. In Fig. 6, I reconstruct the inferential configuration for the argumentative pattern containing the locus from termination and setting up related to the 'garment workers' group and encompassing the categories "Workers being exploited/suffering/abused" and "Workers in a vulnerable condition". Figure 6 shows that it is the same argumentative pattern discussed for 'big brands', with the difference that here the focus is on the negative condition of garment workers, as these are characterization frames of this group of players, and it is not mentioned who is to blame for this situation. On the other hand, in the occurrences of characterization frames of 'big brands', the focus is on them, as the actors responsible for the negative condition of workers.
Through the argumentative pattern illustrated in this section, it emerges that the negative characterization frames employed by activists to refer to big brands' persona and stable behaviour, as well as to garment workers' state/condition, become the cultural shared premise to justify the need for a revolution. In line with the activists' rhetorical goals identified in 3.1, the audience for these arguments is clearly not big brands, but rather a third party that includes both the general public and legislative bodies, e.g. the European Parliament.

Conclusions and Directions for Future Research
At the theoretical level, this paper has advanced the reflection about the relationship between frames and argumentation, by investigating a specific kind of frames, i.e. characterization frames, in association with the argumentative patterns retrieved in a corpus of social media posts published in the context of the public controversy surrounding fashion sustainability. While previous studies in argumentation acknowledged that frames somehow activate underlying argumentative premises, specifically material starting points, this research has made a step forward by showing, through the analysis of argumentative patterns, the dynamics by which characterization frames contribute to the construction of endoxa. Throughout the analysis, the characterization frames associated to the multiple players involved in the controversy have been identified. Drawing on a theoretical framework combining Pragma-dialectics with the Argumentum Model of Topics, different argumentative patterns have been found for each category of characterization frames. A deep investigation of these argumentative patterns has revealed that characterization frames play an argumentative function in the corpus analysed, by constructing two main endoxa that are perfectly in line with the objectives of the Fashion Revolution campaign. The first endoxon that is constructed establishes the essential characteristics and attributes that are needed for claiming to be part of Fashion  Revolution. In respect to garment workers, this means to work with artisans; to have workers being treated fairly; to be supporting workers in a vulnerable condition; to be undertaking practices supportive of workers; to be involved with workers having positive characteristics. For small businesses, being part of Fashion Revolution means to be ethical; to be sustainable; to be focused on the environment; to be small or local; to undertake virtuous practices, in particular treating workers fairly and using sustainable materials; to support workers; to work with ethical partners; to be proud of their sustainable practices; to have a humanitarian goal; to be revolutionary; to be transparent. The second endoxon that is constructed builds a negative evaluation of the practices of big fashion brands, i.e. that the practices undertaken by big brands are negative or, from the perspective of garment workers, that the current state/condition in which garment workers are is bad. This endoxon serves as the basis for justifying a call of action against big brands.
From a methodological perspective, this paper has proposed a refined definition of the concept of characterization frames, elaborating on the original definition offered in conflict resolution studies (see Shmueli, 2008). The refined definition is useful for identifying the discursive representation of these frames in a corpus and is based on the recognition of three constitutive elements of characterization frames: the persona, the stable behaviour and the state or condition of a certain player.
Overall, this study contributes to two different streams of argumentation research. First, it advances studies about linguistics-based argumentative approaches to frames and framing, fostering the reflection about the relationship between frames and underlying implicit premises with a focus on a specific category of frames, i.e. characterization frames. Then, this paper contributes to research about political argumentation, by helping to interpret argumentation in an under-investigated political arena, that is, activists' argumentation on social media. In this respect, this study analyses activists' argumentation on two different social media, i.e. Twitter and Instagram, both important for fashion activism.
This paper represents the first step in a broader research project aimed at investigating the relationship between conflicting characterization frames and implicit argumentative premises in the context of the public controversy surrounding fashion sustainability, with the end goal of offering new analytical tools to better understand and perhaps intervene in this controversy. According to authors in conflict resolution studies (Shmueli et al. 2006;Shmueli 2008), reconstructing the conflicting frames can help to expose the deep beliefs, views and values of the parties involved. In fact, as frames intrinsically offer a simplified interpretation of reality, they enhance the parties' divergent interpretations of the world and thus often lead the discussion to become polarized and exacerbated. In this sense, frames that are consolidated can hinder a party's capacity to see matters more 'objectively' (Shmueli et al. 2006;Shmueli 2008).
As mentioned in the introduction, this public controversy constitutes an instance of complex argumentative polylogue, which involves multiple players, positions and places (Aakhus and Lewiński 2017). While in the present study I focused on the relationship between characterization frames and argumentative patterns in the discourse of one specific player involved in the controversy, that is, activists, in future research I will also consider documents issued by other players, e.g. big brands and public institutions. Since the concept of characterization frames comes originally from conflict resolution studies, in the integration of the new documents I will investigate where the characterization frames used by the different players result conflicting. For example, does the group of players 'garment workers' present the same characterization frames in activists' discourse and in big brands'? Identifying the points where characterization frames appear conflicting between different players and the related endoxa that Fig. 6 General inferential configuration of the argumentative pattern containing locus from termination and setting up-'garment workers' are constructed will allow to gain a better understanding of the argumentative misalignments (see Greco and De Cock 2021) present in the controversy. Furthermore, at the methodological level I plan to further advance the annotation scheme developed in this paper for the analysis of characterization frames and of the related argumentative patterns. This annotation scheme would then be applied to the additional documents considered.
Eventually, in future research it would be interesting to continue the reflection about the argumentative role of characterization frames by considering this research problem from the perspective of another concept developed in argumentation studies, i.e. the notion of argumentative potential (Mohammed 2019). In fact, a study of the argumentative potential contained in different characterization frames could help to understand how they become conflicting when adopted by divergent parties.
Funding Open access funding provided by Università della Svizzera italiana. No funding was received to assist with the preparation of thismanuscript.

Conflict of interest
The authors have no competing interests to declare thatare relevant to the content of this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.