Multimodal critical discourse analysis of the cinematic representation of women as social actors

In this paper, an expanded version of the visual social actor network (van Leeuwen 2008) is suggested in integration with the classifi cation of fi elds proposed by Martin (1992). In the context of multimodal critical discourse analysis, such network has provided a template for the discussion of the data aiming to reveal and interrogate the construal of women’s agency in the introductory part of the fi rst fi lm Sex and the City (2008). Overall results place women’s actions mostly in the private sphere but also reveal instances of gender performativity (Butler 1990) that question social norms for women mainly imposed by the media.


Introduction
In this epigraph, addressing the impact the media have on people's individual and collective imagination, Bauman (2000) draws our attention to possible differences between life on screen and life lived, i.e. between what is represented in telecinematic texts and what is experienced in everyday life. The author seems to allude to diacritic elements in telecinematic discourse (Piazza, Bednarek and Rossi 2011) used as resources to create specifi c representations of reality, which might, in turn, seem more appealing and/or easier to absorb than the real lived experiences of viewers. In this sense, these semiotic resources are "deeply political" (Bouvier 2018:205).
Such claim underlines the imbalance oftentimes created by the media in people's perception of reality and of its socioeconomic consequences in contemporary society, as today's world becomes increasingly more saturated by consumerism and the pursuit of 36.4 2020 identities projecting what seems to be valued socially and culturally (Bauman 2000). In this context, this paper aims to discuss visual representations of women as social actors 2 in the fi rst fi lm Sex and the City 3 (2008) -SATC, thus foregrounding the extant bidirectional relationship between identities and discourse.
The TV series has received considerable attention both in the media (blogs and news outlets) and in academia 4 , in areas such as feminist, media, fi lm and cultural studies. Among these, while Di Mattia (2004) problematizes the search for Mr Right by questioning the idealized male partner as an unreachable model, Montemurro (2004) highlights how the TV series duly depicts the challenging balance between motherhood and work. In the area of language/discourse studies, Bezerra (2008) carries out a critical discourse analysis of gender representation in the discourse of the narrator while Bubel (2011) addresses how friendships are forged.
The results presented in the present paper have been obtained by means of a meaningful, productive and timely blend between studies in multimodality and in critical discourse analysis (CDA) -thus representing a contribution to research developments in multimodal critical discourse analysis (MCDA), a fi eld of inquiry still in its infancy (Djonov and Zhao 2014; Ledin and Machin 2018). Emphasizing issues of language use, power and inequitable relations construed in 2. Aligned with propositions in van Leeuwen (2008), the term 'social actor' is used here to indicate that the analytical emphasis surpasses the linguistic identifi cation of categories of participants involved in the events described to include a sociological aspect of the implications of the modelling construal one builds about/of individuals in (multimodal) texts. Besides, the term 'agency' is recurrently used in this paper to refer to the extent to which social actors may act independently while also being impacted by social, cultural and political structures. Rather than synonymous, these terms should be seen as complementary when observing sociological implications of visual/linguistic representations of individuals in society. 3. This fi lm presents a continuation of the stories in the lives of the four main characters -Carrie, Miranda, Charlotte and Samantha -since the homonymous TV series ended in 2004 after a total of six seasons. 4. Upon search on Portal de Periódicos CAPES/MEC, no papers were found to have investigated the fi lm Sex and the City since its release in 2008. The search criteria were: sex and the city; fi lm; last 10 years; any language; peer-reviewed journals. Whenever 'Sex and the City' was mentioned in the papers found, it was either about the TV series or only mentioning the fi lm (among others) as an instance of the theme being discussed, i.e. it did not make up the data being analyzed. Last access on May 5 th , 2019.
Fábio Alexandre Silva Bezerra and through discourse by a combination of modes in a wide range of contexts, MCDA nonetheless offers a useful template for the discussion of gender-related issues in contemporary society by investigating how multiple semiotic modes create more subtle representations of gender inequity, especially those in popular culture media, such as TV, fi lm and magazines.
In this respect, Ledin and Machin (2018:60) highlight the need for "clear, robust concepts [in multimodality] that can be used as part of CDA with its emphasis on digging out the discourses buried in texts to reveal (…) power relations and ideologies", thus strengthening the connection between these two fi elds of inquiry. MCDA, as an interdisciplinary approach, attempts to pull research out of its disciplinary silos, incorporating contributions of different fi elds for a more robust understanding of how multiple semiotic modes (verbal, visual, haptic, auditory, etc.) interact textually and discursively.
Importantly, besides revealing possibly biased representations and contesting oppressive social relations, the critic also needs "to take up the fair position (…) to not only negate but to add" (Roderick 2016:164). In other words, research in MCDA also needs to question the reasons for such imbalances, as an attempt to start de-naturalizing beliefs that sustain unequal relations of power in society.
Although numerous studies have addressed the ways in which still images construe representations of reality in systemic-functional and/or socio-semiotic terms (Bezerra, Nascimento and Heberle 2010;Bouvier 2018;Chen and Machin 2014;Kerry 2016; among many others), there is considerably more to be achieved in terms of theoretical and methodological consistency as far as dynamic 5 images are concerned. Aiming to add to important contributions worldwide (Bateman 2009;Bezerra 2012Bezerra , 2016Böhlke 2008;Iedema 2001;Lima-Lopes 2016;O'Halloran 2004;Tseng 2009;van Leeuwen 1996), this paper, adopting a combined systemic-functional (Halliday and Matthiessen, 2004;Martin 1992) and socio-semiotic approach (van Leeuwen 2008), proposes to reveal how the representations of women in the fi lm SATC 5. Dynamic multimodal texts are "fi lm and video texts which display different and constantly varying confi gurations of sound, image, gesture, text and language as the text unfolds in time" (O'Halloran 2004:110).
36.4 2020 either expand or limit the domains of action in which they can operate. From Halliday and Matthiessen (2004), the notion of system networks is explored; from Martin (1992), part of the provisional classifi cation of fi elds is used; and from van Leeuwen (2008), the visual social actor network is expanded, thus offering a broader foundation for the discussion of women's agency in this fi lmic production.

A methodological account
The results discussed are part of a wider investigation in the fi eld of MCDA (Bezerra 2012), having been updated considering studies in the area published in major national and international journals in the past six years. While this paper emphasizes the visual elements, my unpublished doctoral research (Bezerra 2012) provides a more encompassing verbal, visual and intermodal analysis of SATC. Additionally, due to constraints of space, only the introductory part (3'50") of the fi lm (Intro) is analyzed here in order to demonstrate and discuss the expanded visual social actor network proposed.
In the Intro, the audience is given a summary of the most important moments in the four main characters' lives throughout the originating homonymous TV series, thus allowing for fans to be reacquainted with the characters and for newcomers to have some background information before seeing the new events in the fi lm unfold.
In the broader investigation (Bezerra 2012), a number of steps had to be taken to obtain the data presented and discussed in the analysis section about the visual representation of women as social actors. First, both the verbal and visual texts were broken down into the smallest units of analysis (clauses and shots, respectively), which also allowed for the intermodal analysis to be carried out. The representational resources mobilized in the multimodal text were analyzed in systemic functional terms (Halliday and Matthiessen 2004), where the ideational content is described as processes, with attending participants in specifi c circumstances.
These processes could be realized as action, perception, cognition, emotion, desideration and saying. The participants involved in such processes in specifi c contexts could be either dynamic or passive.

Fábio Alexandre Silva Bezerra
The fact that the dynamic image, like the verbal text, can also realize processes of emotion and desideration is also proposed. In other words, unlike the still images (Kress and van Leeuwen 2006), the dynamic text presents a greater ideational meaning potential -much in the same terms as verbal language. Having explained that, let us focus on how this specifi c set of data was formulated as only images are addressed in this paper.
Using software for video editing (ELAN and FastStone Player), a detailed layout of the dynamic image in tables was presented, and all actions which women carried out could be informed. In the analysis section ahead, based on Table 2, only the actions with signifi cance in regard to gender identity are listed and discussed. For a more thorough description of the steps in the analysis of the dynamic image, see Bezerra (2012), where the text was separated into comprehensive units proposed by Baldry and Thibault (2005) to be able to identify and classify the data accordingly: whole text > macro-phase > phase > sub-phase > shots > visual transitivity frames.
Following all these levels of analysis for the dynamic image, it was possible to locate and quantify the actions in which women were involved, thus allowing for a discussion of the complexity and extent of their agency by utilizing the expanded visual social actor network presented and explained in the next section.

Some theoretical underpinnings
In order to reveal and discuss the visual representations of women in SATC, emphasis has been placed on how they have been portrayed as social actors and the implications of such construal for women's agency in contemporary society, especially considering that identities are often produced through articulations with popular culture. In that respect, the theoretical underpinnings that guide the concept of gender in the present study are delineated next.
Generally speaking, there are two opposing ways to view and explore identity (Benwell and Stokoe 2006): the essentialist and the constructivist perspectives. On the one hand, (gender) identity can be understood as something essential, pre-discursive and, therefore, 36.4 stable -rendering it as something that one possesses. On the other hand, identity is seen as constructed, discursive and, therefore, unstablemeaning something that one performs.
In line with the discursive view of identity, informing the present study, Butler (1990Butler ( , 2004 describes gender as performativity. This concept purveys the notion that gender is something that one does, something that is actualized in the everyday acts of performativity, entailing a reiterative practice (Roderick 2016), rather than a pre-given construct. A number of studies have addressed how performativity is construed by tackling social issues like weight loss (Mishra 2017) and masculinities (Kerry 2016;Pimenta and Natividade 2012), among others.
Understanding gender as performativity contributes to dessentializing notions of identity, avoiding generalizing interpretations of individual and collective experiences. In other words, women's social agency should be understood as malleable, which does not mean believing in the chimera of unrestrained freedom to experience and perform identity in all contexts, as individuals and different social groups are mobilized by the social structures under which we all operate. A multimodal critical discourse investigation of gender representation ought, therefore, to address how images construe identity by deploying the semiotic potential available to meet particular discursive goals in specifi c contexts.
The relevance of the focus on the meaning potential of the image has been recently stressed by some scholars arguing for a socially and affordance-driven approach to MCDA (Machin 2016;Machin, Caldas-Coulthard and Milani 2016). Such approach is informed by the work of Halliday (1978) and Halliday and Matthiessen (2004) in systemic functional theory (SFL), emphasizing "not the idea of a formal, strict grammar but one of a system of semantic choices and networks" (Machin, Caldas-Coulthard and Milani 2016:302).
Although an important aspect of an affordance-led approach to MCDA is to describe and discuss the meaning potential of the specifi c semiotic resources under analysis (see Bezerra 2012), the core aims of this paper, as previously indicated, are to describe and expand a visual social actor network as well as to discuss and problematize the construal of women's agency in SATC.
In the fi rst heuristic explorations of the data using the network proposed by van Leeuwen (2008), it became apparent that more needed to be said about the representations found in the analysis than the system network available seemed to offer as described options of meaning. Importantly, this expansion is surely not intended as criticism towards the originating network (van Leeuwen 2008); conversely, its aim is simply to highlight to the analyst the more in-depth analysis that may be carried out by moving rightwards in the system previously described as additional levels of delicacy were proposed.
This path (from data to system network) allowed for an interpretation of the results that was not simply limited to the network proposed and its terminology, thus avoiding the case that "examples are chosen that allow concepts and models to be illustrated" (Machin 2016:326, citing Forceville 1999. Instead, the system network was expanded so as to account for the aims of the present study, based on the premise that "the actions and agency of represented actors can be analysed by attending to exactly what types of activities they perform" (Roderick 2016:397).
Since the main concern here is to address visual representations of women in SATC in regard to their spaces for action and the impact they might have on other people or things, van Leeuwen's (2008) visual social network ( Fig. 1) proves useful in describing the goings-on in the dynamic image in terms of what women are (not) represented to do and with whom they interact when doing these things.

36.4
In this system network, van Leeuwen describes options of meaning when considering the potential available for representing social action. First, individuals may be either included or excluded from the visual text. If included, there are three additional sub-sets of possible meanings considering their involvement in actions as agents or patients, their categorizations and their placement as individuals or groups. For the purpose of this paper, only the sub-set of involvement in action is considered, which is where the expansions proposed appear.
The fi rst expansion consists in the addition of considerations about the fi elds of discourse in which the actions involving agents and patients are located. In other words, aiming to address women's social agency in SATC, it did not suffi ce to simply describe the actions with which they are involved; equally important was to classify these actions according to the contexts in which they are found. For that, the classifi cation of fi elds proposed by Martin (1992) was partly integrated with van Leeuwen's (2008) visual social actor network. Before explaining how this integration occurs, a brief description of Martin's system network below is provided.
Discussing field agnation, more specifically the nature of socialization in such fi elds of discourse, Martin (1992:544) fi rst describes two main entry points into the system (Fig. 2): 'oral transmission' and  'written transmission'. In all the data analyzed, only one case of 'written transmission' was found, when the character Miranda is represented reading and signing a contract to become a partner in the law fi rm where she works. For this reason, only the feature of 'oral transmission' was considered in the analysis presented in this paper. Looking at the system 'oral transmission', the feature 'domestic' (renamed here as 'non-specialized') is in contrast with the specialized fi elds of discourse. In these, one may be involved in recreational or trade events, with the former being further divided into activities related to sports and hobbies. As the primary focus in this paper is to state if women's actions are placed in the specialized or non-specialized fi eld, the further sub-features of specialization were not considered.
In regard to the accuracy of such classifi cation of fi elds, Martin (1992:546) cautions against "the danger that the more domestic fi elds be misinterpreted as natural theories of experience with the more technical ones treated as semiotic construals of reality". In an effort to avoid statements that might place either women or men in pre-established roles pertaining to conceptions of what is perceived to be natural or cultural, such claim is rather signifi cant considering that "all fi elds are semiotically constructed" (Martin 1992:546).
Besides adding the features 'specialized' and 'non-specialized' in the sub-system of 'inclusion' (van Leeuwen 2008), the expansion proposed here also adds more levels of delicacy to the sub-feature of 'agent'. For the fi nal expanded system network, see Figure 3 below. First, notice that the agent may or may not affect someone or something. If affecting, they might affect either themselves or others. If affecting others, these may be either people or things. Now, if not affecting anyone or anything, the agent may only be acting (e.g. walking) or behaving (e.g. laughing). These additional levels of delicacy arguably allow for a more in-depth look into the types of actions and their reach, which is rather useful when discussing the complexity and extent of women's agency in the fi lm under analysis.
In the following section, this system network, more specifi cally the feature 'involved in action' along with its sub-features and corresponding levels of delicacy, is explored in the analysis of the data collected from the Intro of the fi lm, aiming to reveal and hopefully contribute to de-naturalizing the ways in which women's agency have been portrayed, which arguably informs how they are perceived by society at large.

The visual representation of women as social actors
Having outlined the expansion proposed to van Leeuwen's (2008) visual social actor network, and stressed the focus on the discussion of gender issues as far as women's agency is concerned, next is description and discussion of the data found in the Intro of the fi lm regarding the dynamic image only.
When discussing how participants are represented visually, one fi rst needs to consider what has been included in and excluded from the multimodal text (Fig. 3). First, the discussion emphasizes who has been excluded from the depictions precisely because it "is always a symbolic form of social exclusion, not acknowledging the existence of certain people or kinds of people who live and work among us" (van Leeuwen 2008:142). Roderick (2016:396) also adds that, besides "analysing how actors are represented, it is equally important to consider those meaningful absences which equally underwrite the homogeneity/ solidarity being produced through categorisation" in the visual text.
The fi rst noticeable absence from the representations in the Intro is that of black people -not being displayed even in the background, much less interacting with any of the four main characters. Highlighting this exclusion is important especially because the introductory part of the fi lm, as previously explained, presents a summary of the most relevant events and character relations in the TV series, thus revealing that both the TV series and the fi lm are almost entirely about white people living in New York (Manhattan and Brooklyn). This fact revealed from data analysis is especially relevant as "gender imbalances never operate alone but intersect with other axes of inequality" (Machin, Caldas-Coulthard and Milani 2016:306), such as class, age, sexual orientation and ethnicity.
Before further analysis of the Intro, the character Louise, Carrie's personal assistant in the fi lm, needs to be mentioned, as her role, though minor, is not at all insignifi cant. However, she only appears later in the storyline. Since this paper focuses on the fi lm Intro due to its relevance, the fact that no black person is displayed in any of the frames is in itself revealing of a lack of racial/ethnical diversity.
Other important markers of difference almost entirely absent from the depictions are those of older age and heavier body size. In general, depicted individuals are young, slim, white, apparently middle class, straight women and men. Although this is a signifi cant result, the character Harry (Charlotte's second husband), though appearing rather briefl y in the visual text, is an important representation of diverse body size and image. Having fallen in love with Harry, Charlotte even converted to Judaism, his faith -which, opportunely, is also duly represented in the dynamic image analyzed, thus refl ecting an instance of attention to religious diversity.
Such lack of representativity seems markedly less realistic knowing all events occur in New York -one of the most diverse cities the world over in numerous aspects. Statistically speaking, for instance, the Organisation for Economic Co-operation and Development (OECD 2017) reports that 38.2% of the adult population in the US is obese, of which women's percentage is a little over that of men (41% women, 35.5% men). This is why a multimodal analysis of gender identities needs "to think about the way that people in representations are placed in the social world, which parts of their identity are foregrounded, backgrounded or hidden" (Chen and Machin 2014:290).
Regarding sexual orientation, no identifi able gay/lesbian/bisexual characters are displayed in the Intro, with the exception of two male extras who kiss in the background (much to Samantha's disappointment). There are, nevertheless, two gay men who appear later in the fi lm, and who also were characters in the TV series: Stanford and Anthony -Carrie's and Charlotte's best male friends, respectively. Precisely for being best friends with two of the main characters, their absence in the Intro cannot be overlooked -leaving the audience to wonder why they are not placed more centrally in the fi lm as well.
Although these are not the only absences in the visual construal of the Intro in this fi lm, underlining ethnicity, age, class, sexual orientation and body size seems suffi cient to clarify who is primarily depicted as viable images in the telecinematic text under analysis. Next, considering the choices of inclusion in the data analyzed, results might contribute to understanding the constraints and the potential of the dynamic image for more equitable (gender) representations.

Fábio Alexandre Silva Bezerra
In order to understand the annotations in Table 2, see below the abbreviations of the names of the fi lm characters appearing in the shots analyzed in the Intro (Table 1). For instance, when referring to the visual occurrences of women getting married in the Intro, Table  2 shows "getting married [TR, HR, ST]", which indicates the two instances where Charlotte gets married, fi rst to Tray (TR) and later to Harry (HR) -and then the portrayal of Miranda getting married to Steve (ST). In other words, the abbreviations in Table 1 show the people or things that are the recipients of women's actions.
These are not, however, the only people depicted in the dynamic image. Other individuals also form the background against which the protagonists and their co-stars interact. Some of those extras are at times mentioned whenever their visual construal is noticeable enough to allow for analytical work as far as gender issues are concerned, being referred to in general terms as woman (W1 and W2) or man (M1) -the numbers indicating the order in which they appear on the screen.  The fi rst more general observation is the absence of actions in which women affect themselves, underscoring the focus of their agency on either other people or things (affecting), as well as on actions which are intransitive (unaffecting). This result indicates the decision in the Intro to present the four main characters by emphasizing their actions and the people with whom they interacted throughout the TV series.
The second more general observation is that there are more instances of women's social agency not affecting anyone or anything. They are depicted visually 63 times either acting (39) or behaving (24) as opposed to 35 instances affecting either people (24) or things (11). This fact seems to indicate that women are mostly construed to be perceived as individuals, focusing on their personal characteristics. Such decision is arguably aligned with the main objective of the Intro to (re)introduce the main characters to viewers. Importantly, all 24 instances of women affecting other people are limited to their personal relationships, which possibly circumscribes performativity to the private sphere. As Butler (1990) argues, (gender) identity is actualized in specifi c quotidian actions, part of reiterative practices creating the general idea of how individuals are to be perceived. In this case, the construal seems to position women in places where their interactions only seem valid in private, more personal situations -a disquieting notion, considering the real pervasive impact of women in today's contemporary society in a wide range of contexts and domains.
Besides being restricted to the characters' private lives, the 24 occurrences of women's actions affecting other people are circumscribed to fi ve rather specifi c actions: getting married, hugging, having sex, kissing and holding -arguably implying that these actions constitute the more substantial aspects of their lives. Even taking only their private lives into account, such a constraining range for women's actions needs to be questioned and rejected for being utterly unrealistic and problematic.
Another striking result in terms of (gender) identity representation is the portrayal of Samantha, in another montage, having sex with seven different men on different occasions (Fig. 4) 6 . With sex being a pivotal element of both the TV series and the fi lm, why not include instances of the other characters having sex, thus creating a more truthful construal of the events in the TV series? Although Samantha is fi ttingly acknowledged for her free-thinking ideas and practice of sex, having all 7 instances focusing solely on her does not seem to be justifi able. Having said that, one must also remember that only a very small cohort of women over forty years old are represented as sexually active in telecinematic discourse (Piazza, Bednarek and Rossi 2011), making Samantha a meaningful example in that sense.
Additionally, no other character (female or male) is depicted interacting with people other than their friends and family, except for one situation in which all four main characters (C, CH, MR and SM) smile at an attractive young man passing by only to fi nd out he is gay as he kisses another man waiting for him in the background (Fig. 5). Seemingly innocuous, this result seems to place most of women's agency in their private/social spheres in interaction with only other familiar men/women.  6. Common practice in the publishing sector, a limited portion of copyrighted material may be used when constituting fair use. According to Section 107 of the U.S. Copyright Act, "the fair use of a copyrighted work, including such use by reproduction in copies or phonorecords or by any other means specifi ed by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright" (USCO 2016:19). Nevertheless, full attribution must be given to HBO -Home Box Offi ce, Inc. as the copyright holder of the motion picture from which the still images in Figures 4-13 were extracted for use in this paper.
Emphasizing the instances of performativity of motherhood, there are depictions of Charlotte (Fig. 6) and of Miranda (Fig. 7) holding their children, creating the image of dedicated caretakers, both in their family leisure time accompanied by their husbands (Harry and Steve, respectively). Unsurprisingly, the focus on Charlotte's motherhood is present throughout different shots, since this is a rather important aspect in her life having been unable to conceive, leading her and her husband to adopt Rose.
However, having chosen her career over early motherhood, Miranda had a signifi cantly different experience as she got pregnant unexpectedly, marrying Steve sooner than planned. The scene of their wedding ceremony is where one of the two instances of kissing in the Intro appears, when bride and groom fi nally kiss.
The other instance occurs when Carrie kisses Mr. Big (Fig. 8) in front of his apartment building. Considering that a substantial part of the stories shared by the four main characters in the TV series was about their personal (sexual) relationships with other men, the Intro surprisingly emphasizes only Miranda's and Charlotte's latest relationships -showing their husbands and children, while creating a better summary of events for Carrie (previous boyfriends) and Samantha (previous sexual partners). Looking at the instances where women acted onto things, most of them are related to their social range of action (drinking, putting on make-up/sunglasses, eating), when the main characters are depicted in restaurants/bars having a meal and drinking with each other, construing the concept of close-knit friendships. The other 3 instances show Carrie holding and looking at a photo of herself and her three best friends, and holding the books she wrote, portraying an action with a broader social scope. However, instead of showing Carrie at a bookstore, for instance, where her books would be in display, the depiction was simply of her holding the books in her bedroom. This way, an action with a more public relevance still fi nds itself somewhat confi ned to a private domain. Now, regarding occurrences of women's agency not affecting anyone/anything, they describe two types of events: women are either acting (39) or behaving (24). When acting, they are mostly depicted either walking (18) or standing (7), with other actions happening less frequently: working (6), lying down (3), screaming (3) and running (2). The data results show women mostly walking around the city, which helps create the context in which their lives unfold: New York -one of the most recognizable, ethnically diverse and densely populated cities in the world.
Walking around the city, these women also highlight other elements of life in the Big Apple that have been an integral part of SATC: commerce, fashion and relationships. For example, in the opening frames of the Intro, there is a young woman (W1) depicted standing in front of a shop window and smiling at a mannequin (Fig. 9), revealing her interest in what is for sale, confi rmed by the accompanying narration: Year after year, twenty-something women come to New York City in search of the two L's: labels and love. The role of fashion in SATC is mostly defi ned by wearing designer clothes and shoes, privileging a desirable image contradictorily unattainable for most women.
In addition, Carrie is represented walking and smiling (Fig. 10) as she passes by a group of young girls resembling herself and her best friends. In this shot, Carrie is wearing a white gold fl ower cocktail dress, aiming to set her apart as the fashion-forward leading character.  Relationships are another key element in the visual construals of SATC, whether referring to physical ones (Fig. 4), to friendships (Fig. 5) or to romance (Fig. 6, 7 and 8). Even though different types of relationships are depicted, it appears that women are acknowledged as sexual beings only privately, except for Samantha, the overtly more sexual character, who is more often than not criticized for an allegedly lack of moral probity. Such constraining depictions in the media need to be questioned, as they seem to legitimate narrow perceptions of gender, limiting them to roles and behaviors, when it needs to be understood as a complex web of factors that intersect in one's identity.
The most frequent occurrence of women's actions not affecting people or things is that of working (6), showing the main characters (except for Charlotte) involved in actions with a more public scope. However, out of these instances, only one actually happens outside of their home, when Miranda (a lawyer) can be seen in her offi ce (Fig.  11). Another instance shows Samantha (a PR executive) working from home (WFH) while bathing in the sun in her bathing suit (Fig. 12) alongside her boyfriend (and client) Smith.
In the other four instances, Carrie is using a computer to write for her newspaper column from home as well (Fig. 13). Her work has been construed visually by means of a montage, where four different shots (at different points in time) are placed together in the same sequence of frames. This representation seems to purvey the idea that being a professional writer is a signifi cant aspect in Carrie's life, thus creating the reiterative factor in the performativity of her (gender) identity. Although such result could be at fi rst interpreted negatively as women are not seen occupying the offi ce workspace, both Samantha and Carrie are self-employed, allowing them to decide their place of work. Additionally, this greater presence of women working from home seems consistent with a recent study by the United States Department of Labor (USDL 2016) showing that more women (23.5%) work from home than men (21.5%) in the USA, yet not necessarily receiving remuneration commensurate with their male work counterparts.
Quite oddly, in all instances of behavior depicted, women are simply smiling (24), which arguably helps to sustain an enduring connection between women and emotions (Galasinksi 2004). Such imbalances need to be examined as "hegemony relies on our perception that the dominant messages are just the way things are" (Kerry 2016:232).
At last, Table 2 shows all fi ve female characters depicted (CH, MR, SM, C and RS) and one extra (W2) as patients in only two types of actions: the acts of hugging (4) and holding (2). In all occurrences in which women are being hugged, the agentive position is occupied by men with whom the women are romantically involved: M1/W2, HR/CH, ST/MR, SH/SM and MB/C. This result seems to confi rm the choice to construe only heteronormative relationships in the Intro, even though gay, lesbian and bisexual relationships did exist in the TV series. For instance, Samantha dated Maria (played by Brazilian actress Sônia Braga) in the fourth season in 2001.
Whereas such representations seem to constrain ideas and images about romantic relationships in contemporary society as far as sexual diversity is concerned, the stable relationships of two of the four main 36.4 2020 characters seem to diverge from the social norm in relevant ways: both Miranda and Samantha date younger men who are less educated and rather less fi nancially stable.
Miranda eventually marries Steve, while Samantha and Smith decide to live together in Los Angeles, where he could pursue his acting and modeling career more successfully. With their partners' support, both Steve and Smith end up fi nding a successful career path, underlining the positive impact their wives had in their lives in opposition to other frequent media representations of women's relationships where they are oftentimes markedly more pliant to the desires and needs of their male partners, especially when fi nancially dependent.
In the expanded visual social actor network (Fig. 3), besides analyzing women's involvement in actions either as agents or patients, one must also consider in which fi eld of discourse their actions are mostly located. In that respect, another salient result is that only one type of action has to do with the specialized fi eld: that of working in its six occurrences, as previously described. All other instances are situated within the non-specialized fi eld (Martin 1992), more specifi cally that of personal and romantic relationships, thus revealing the construal of a limited scope for women's agency.
Unlike what is depicted in the Intro, our lived everyday experiences confi rm that women occupy varied areas of social life, impacting signifi cantly on different groups and individuals. Such observation is pertinent considering that "the visual is particularly important in global communication since it can claim authenticity and resemblance, rather than being simply opinion" (Bouvier 2018:194), thus stressing the importance of critiquing media representations in order to interrogate the extent of their reliability. As (gender) identity is construed by means of performativity of reiterative actions in our daily experiences (Butler 1990), these representations of women in limited areas of life need to be called into question.
Even though the network proposed as analytical tool in this paper (Fig. 3) offers the critic a template from which to draw conclusions and critique imbalanced power relations and biased representations, it is not meant as an exhaustive description of possible meanings regarding 36.4 2020 Fábio Alexandre Silva Bezerra social agency. Much like Sunderland and McGlashan (2013:492), we hope to have been able to demonstrate that meanings are "always contingent, co-constructed, interpretive, and cannot simply be read off from the [dynamic images] via the frameworks" proposed for analytical purposes.
Having described and discussed the construal of women's agency in the Intro of the fi lm SATC, drawing data from the dynamic images by applying the expanded visual social actor network (Fig. 3), the next section presents a summary of the main fi ndings as far as (gender) identity is concerned. Additionally, the relevance of conceptual and methodological proposals in MCDA is underlined for a broad spectrum of multidisciplinary studies, also stressing its signifi cance for pedagogical purposes.

Concluding remarks
The main objectives of this paper were twofold: 1) to introduce and use the expanded visual social actor network (Fig. 3), and 2) to reveal and discuss the construal of women's agency in the Intro of the fi lm insofar as the performativity of (gender) identities was concerned. The study was conducted within an MCDA approach in order to address how multiple semiotic modes can construe depictions that are oftentimes pervaded by inequitable power relations, setting forth biased gender representations in the media. Equally, attention must be paid to the constructive and emancipating power of discourse concerning its capabilities in creating more evenhanded social relations and identity portrayals.
Identity is a performative endeavor constituted in the discursive reiterative practices of our everyday lived experiences (Butler 1990). Therefore, the construal of women's agency in the fi lm was discussed highlighting how the varied representations diffused in society impact the way we perceive ourselves and behave socially. Understanding that media representations possibly become "a form of one's social and subjective identity" (de Lauretis 1999:260), there needs to be a move beyond restrictive notions of gender roles and differences to include opportunities for the depiction of people's identities in their rich complexities and intersectional relations of gender, race, ethnicity, class, age, spirituality, body diversity and sexuality.
Regarding the main results, almost all instances in the data concerned the non-specialized fi eld of discourse, placing women's action mostly in the private realms of society -with the sole exception being the scarce depictions of three of the four main characters working. Second, of the actions in which women affected others, most referred to their intimate relationships with their partners (e.g. hugging, having sex, kissing) and their close relationships with their friends (e.g. drinking, eating). When their agency did not affect anyone/anything, most had to do with them just standing or walking around the city, and most strikingly with them simply smiling.
Despite the fact that we are all increasingly more "aware that 'belonging' and 'identity' are not cut in rock, that they are not secured by a lifelong guarantee, that they are eminently negotiable and revocable" (Bauman 2004:11), the visual construals in the Intro seem to create a generalized notion of women as more emotional and mostly inhabiting the private space in relationships that are limited to their partners and closest friends. Despite being a more provocative series when compared to its predecessors, especially in terms of the issues discussed on primetime TV, one could argue that SATC still seems to reiterate a performativity of womanhood rooted in regulatory practices of body weight, age, appearance and race/ethnicity. An encouraging result, however, is that the discourse of the dynamic image in the Intro also creates some space for women to exist outside the generally restricting scope of media representations, which clearly demonstrates that power in discourse and social processes may function towards more equitable representations. For instance, in times when human bonds seem to be increasingly frailer (Bauman 2004), seeing the four main characters in enduring friendships with one another is quite signifi cant.
Additionally, even though marriage and motherhood are given prominent attention in the Intro of the fi lm, one of the four main characters decides to break up her relationship in search of more fulfi lling personal and professional experiences. While appreciating the legitimacy of marriage, being single can also be construed as a valid choice for women in society, thus expanding the possibilities for identity work to be expressed socially.

Fábio Alexandre Silva Bezerra
The fi lm Intro also seems to present more equitable representations of motherhood, as varied confi gurations are depicted. While Carrie and Samantha are single with no children, Charlotte and Miranda are married with one child each, having gone through quite different experiences. While Charlotte could not conceive and decided to adopt a baby with her husband (Harry), Miranda got pregnant by accident with her boyfriend (Steve), with whom she gets married. Such different experiences portrayed contribute to making visible the many ways motherhood (and fatherhood) can be experienced in society, stressing that the perception and construal of one's gender is a matter of performativity (Butler 1990(Butler , 2004 as they are reiterated in our social and cultural exchanges, i.e. they are not instinctual. In general terms, results indicate a placement of women's agency mostly in the private realm, with only a few depictions falling outside the socially and culturally regulatory norms. These representations are, in nuce, restricting but also enabling, making it ever more imperative to investigate and interrogate the ways in which the media may impact our beliefs and decisions, thus offering us ideas about reality. Therefore, collective actions towards more responsible and engaged participation in our everyday social practices is paramount, potentially enabling more balanced (gender) identity relations in society.
Notably, in the present study, MCDA has proved to be an approach to multimodal discourse that allowed for a discussion of how varied semiotic resources can be used to construe specifi c representations in the media. Such perspective is arguably applicable not only in language studies but also in other knowledge domains, such as sociology, cultural studies, feminist and media studies, since it emphasizes how meanings are created multimodally to sustain or challenge unequal power relations through discourse.
Finally, critically oriented multimodal studies also have important pedagogical implications, considering that school curricula have been increasingly directing "explicit pedagogic attention (…) to the role of images in creating evaluative stance" (Unsworth 2015:73) in multimodal texts distributed across diverse communication channels and by ever-changing technological resources. In this respect, aiming to develop students' multimodal communicative competence (Royce 2007;Heberle 2010), there needs to be a commitment to fostering educational practices that enable teachers to expand their own theoretical and methodological repertoires to deal with multiple semiotic resources and their interrelations in multimodal texts in the classroom. Also, the choice of multimodal texts for the classroom needs to consider a more encompassing view of which literacies can integrate the curriculum, creating a prolifi c link between literacies at school and elsewhere, such as fi lms, TV shows, social networks, and blogs.
For its more comprehensive approach to the study of multiple semiotic modes, MCDA might be the spur schools needed to bridge the gap between what students study at school and their personal interests, thus enhancing the connection between formal learning and their daily lived experiences as readers and writers outside the classroom. In the context of multiliteracies, MCDA may also allow teachers and students to read texts more critically and potentially help to reconfi gure unjust social structures in their lives and in the lives of those around them, contributing to the construction of more fair-minded, unbiased and prosperous collective experiences in contemporary society.