A multimodal analysis of the representation of the Rohingya crisis in BBC’s Burma with Simon Reeve (2018): Integrating Audience Research in Multimodal Critical Discourse Studies


 The paper proposes the integration of Audience Research in Multimodal Critical Discourse Studies to investigate media discourse. The novel approach is exemplified through a case study: the media interaction between one viewer and the text of his choice, BBC’s travel documentary Burma with Simon Reeve. The qualitative research design incorporates a pre- and after-viewing questionnaire, an interview and the multimodal textual analysis. The analytical focus is on the representation of social actors and it is argued that, through this integrated approach, it is possible to explore media effects on an audience by using the cognitive concepts of contextual effects, adapted from Relevance Theory, and ideological effects, a concept derived from the former and modelled around van Dijk’s definition of ideology.


Introduction
This paper discusses a novel methodology integrating Audience Research (henceforth AR) in Multimodal Critical Discourse Studies (henceforth MCDS) to investigate media effects. The methodology is exemplified through a case study that analyses how the BBC travel documentary Burma with Simon Reeve (2018) represents the actors involved in the on-going 'Rohingya crisis', and how one viewer negotiates his interpretation of such actors. The approach combines theoretical and methodological insights from MCDS Machin 2018a, 2018b;Machin 2013Machin , 2016Machin and Mayr 2012), from the social-semiotic approach to multimodality (Kress 2010;Kress and van Leeuwen 1996, 2001van Leeuwen 2005) and from AR (e.g. Schrøder et al. 2003).
Although some reception studies exist within critical discourse research (see further below), recent collections of approaches and methodologies in Critical Discourse Studies (henceforth CDS) show that interpretation from people other than the analyst is still fairly neglected (Flowerdew and Richardson 2018;Hart and Cap 2014). Phelan states that "audience analysis has been relatively invisible in CDA research" and advocates "the emergence of audience-based discourse studies where, instead of starting with given media texts, researchers begin with an analysis of how media audiences have been discursively constituted", with the further aim of "call [ing] into question the residual behaviourism embedded in the image of individual media texts having 'effects' on audiences." (2018, p. 291). The methodology and findings from the case study presented here partly addresses Phelan's call.
The participant was put into an agentive position: he was free to watch a travel or cultural documentary of his choice in whichever way he preferred. He completed a qualitative questionnaire with pre-and post-viewing questions, followed by an interview to explore how he negotiated his ideological positions vis-à-vis the media text. In other words, one might say that this methodology allows to analyse how individuals discursively constitute themselves. However, notwithstanding the participant's freedom of choice, the findings show that text did have 'effects' on him. My approach adapts Sperber and Wilson's notion of contextual effects (1995) and proposes the idea of ideological effects 1 to explore such effects.
2 Integrating audience research in multimodal critical discourse studies

Theoretical and methodological background
A great deal of knowledge of and opinions on events, places and people outside our direct experiential reach are created through mass-mediated discourses realised multisemiotically (Fairclough 2000, p. 165). If we then adopt the critical stance that all discourses, including multimodal ones, are ideological and maintain or resist power relationships (Wodak and Meyer 2016, p. 2), MCDS aptly lends itself to explore the role travel and cultural documentaries play in the hegemonic struggle to conquer the ideological terrain. Although the socialsemiotic approach to multimodality proposed by Kress and van Leeuwen considers the ideological dimension of multimodal discourse (e.g. 2001, p. 34), Machin and Mayr (2012) is arguably the first publication to integrate a multimodal approach within the field of CDS by "seek [ing] to 'denaturalise' representations on other modes of communication" (ibid, p. 9). However, proposals for how to move the field of MCDS forward, although recognising the need to engage with text producers to explore ethnographically how semiotic resources are used (Ledin and Machin 2018a, p. 63), still do not consider audiences as one of the analytical focuses or one of its directions (Djonov and Zhao 2013, pp. 2-3). Audience Research can be defined as "the empirical study of the social production of meaning in people's encounter with media discourses" (Schrøder et al. 2003, p. 147), a task achieved through qualitative methods such as in-depth interviews and open-ended questionnaires. Its basic tenet is that "meaning is never just transferred from the media to their audiences" (ibid, p. 122), which is in line with the non-deterministic stance shared by CDS with regard to the relationship between discourse and society (Wodak and Meyer 2016, p. 20). AR was initially carried out with the issue of media effects in mind. By the 1980s scholars in media and cultural studies had reached three main conclusions: 1) media are only one of the variables influencing audiences; 2) media influence is itself dependent on other facilitating factors; 3) different groups and individuals may react differently to the same media message (Blumler andGurevitch 1982/2005, p. 234). Subsequent work attempted to unpack those conclusions. Starting with Hall's encoding/decoding model (1980Hall's encoding/decoding model ( /2005, AR-based media and cultural studies shifted the focus from the text to the recipients through concepts such as Fiske's polysemy (1986), that is the idea that the texts themselves are open to more than one interpretation; Condit's concept of polyvalence (1991), that is the idea that the denotative message of a text is unambiguously received by the audience who then applies unique connotative filters; Fish's idea that a text only has meaning once interpreted by an audience and his concept of interpretative communities (1980), which sees text producers and receivers as one entity.
Epistemologically, AR can be associated with social constructivism as it sees empirical data as a discursive construction emerging from the interaction between researcher and participants (Schrøder et al. 2003, p. 147). AR and MCDS are thus compatible from an epistemological point of view and integrating a focus on the audience provides a number of advantages for MCDS both from a theoretical and methodological point of view. From a theoretical point of view, being able to triangulate the analyst's interpretation with that of an audience reflects, better than text analysis alone, the non-deterministic stance of the critical practitioner. That is, it gives the opportunity to explore how discourse and society interact by taking into consideration not only the potential meanings found in texts, but also the actual readings of those who interact with them. This also aligns with Hall's non-deterministic view of the relationship between media and audiences and his categorisation of dominant, oppositional and negotiated codes as possible positions at the receiving end of a mediated interaction (1980/2005). A dominant code entails viewers recognising the message conveyed and accepting its validity; an oppositional code sees viewers recognising the message, but deciding to reject it for not fitting in with their worldview; a negotiated code involves viewers accepting the message under certain circumstances (e.g. at a global or national level), but not under others (e.g. within the specificity of their lives), hence negotiating its validity.
Some CDS practitioners have used reception methods in what I call recontextualization studies. Recontextualization refers to the way "texts […] move between spatially and temporally different contexts, and are subject to transformations whose nature depends upon relationships and differences between such contexts" (Wodak and Fairclough 2010, p. 22). These studies involve showing a text to a group of participants and then investigating their reactions through how they recount and evaluate (i.e. recontextualize) what they saw (Benwell 2007;Chouliaraki 2000Chouliaraki , 2003Edward 2016;Kalyango 2011;Kosetzi 2007;Paterson, Coffey-Glover and Peplow 2016;Phillips 2000;Richardson 1994Richardson , 2000. These studies, however, used texts chosen by the researcher, thus making it difficult to establish any prior interest or whether different interpretations were driven by the text or already existing ideologies. A second strand of CDS research, which I call engagement studies, partly addressed the methodological shortcomings of the recontextualization studies. Here participants had spontaneously engaged with the texts chosen by the analysts, thus showing a genuine interest in the text (e.g. Angouri and Wodak 2014;Demetriou 2019;He 2019). These studies, however, tend to be based on comments left by readers of online newspapers or other fora and could still not investigate the extent to which the text had an effect on the audience. Overall, the CDS reception studies validated two main theories in media communication, Hall's model, in so far as the participants did or did not align with the preferred readings identified by the analysts, and Condit's concept of polyvalence, in so far as the participants unambiguously understood the denotative message of the texts but applied their own connotative filter by expressing different opinions. A full-blown AR methodology with participants as agentive subjects and qualitative instruments to investigate the media interaction in-depth, is what I propose here to investigate the thorny issue of media effects.
Furthermore, the proposed methodology can reduce the researcher's bias at two stages: at the point of choosing the text and at the point of evaluating the analytical methods applied for the text interpretation, which are criticisms that have been raised against critical approaches to language and multimodal research alike (e.g. Forceville 1999;Stubbs 1997;Widdowson 1995). However, integrating such an approach entails 'interpreting interpretations', a process dubbed double hermeneutics (Giddens 1979), whereby the researcher "interprets a lived reality that has always already been interpreted by the (senders and) receivers of media, and which may be reinterpreted through the intervention of research" (Jensen 2019, p. 143). Therefore, one has to be careful with issues connected to the 'observer paradox', that is the possibility that participants act or behave in the way the researcher wants them to (Labov 1972), at two stages: the participant's text selection and the participant-generated data. The research design and instruments presented here attempt to minimise this effect at text selection stage by exploring the motivations behind the choice of the programme in the previewing part of the questionnaire and during the interview. Likewise, during the interview the focus is on the topics and actors the participant highlighted in the post-viewing part of the questionnaire and attention is paid during the analysis to those topics or actors spontaneously discussed by the participant, a methodological choice also made by others (e.g. Lotz 2000, cited in Lotz andRoss 2004, p. 504). These measures, combined with the fact that the researcher is not present before, during and after the viewing experience, are in place to minimise issues related to the observer paradox.
Finally, AR allows to analyse media interactions in their individuality. This is not a minor point to consider if we are to reach a better understanding of how media discourse works in shaping society whilst acknowledging that different individuals may have different interpretative experiences (Whitehouse-Hart 2014, p. 5). Research on multimodality with a focus on reception suggests that interpretations differ depending on a number of individual factors: "the task or goal of the [text] examination, previous knowledge and expertise, expectations, emotions and attitudes. Apart from viewer characteristics, even the context in which [texts] are displayed, perceived and interpreted plays a role" (Holsanova 2014, p. 340). A first generation of studies using eye-tracking and/or other psychophysiological measures (e.g. heart rate, skin conductance, facial electromyography) highlighted that multimodal interpretation occurs through the interplay between bottom-up (i.e. text-driven) and top-down (i.e. participant-driven) processes (e.g. Bucher and Schumacher 2006;Chua, Boland and Nisbett 2005;Holsanova et al. 2005). A second generation of studies integrated psychophysiological measures with qualitative research instruments (e.g. retrospective interviews, think-aloud protocols and knowledge tests) to triangulate their data and provide fuller explanations of how multimodal texts are interpreted (e.g. Bucher and Niemann 2012;Gidlöf et al. 2012;Müller et al. 2012). This second generation of studies showed interpretative variance based on the factors mentioned above, urging some to talk about multimodal discourse interpretation as a dynamic and nonlinear phenomenon (Bucher 2017, pp. 94-7). This is particularly true with regard to how a text fits people's existing ideological milieu and which aspects of the text may go to confirm or reject assumptions therein. Although it may not possible to investigate whether ideologies in a specific text have a long-lasting effect on an audience with one qualitative investigation, the analytical concept of ideological effect can be used to at least establish what an audience has focused on and found relevant vis-à-vis their existing ideology.
The idea of ideological effect is derived from Sperber and Wilson's notion of contextual effect in their Relevance Theory (1995). Contextual effects represent the alteration of one's cognitive environment through a communicative exchange. They emerge from the interaction of old and new information, based on two factors: the extent to which the new information modifies or improves 2 the old one and the extent to which the effort required to process the new information is small. Hence, the greater the modification or improvement and the smaller the effort, the greater the extent of the contextual effect. Moreover, contextual effect can come in two types: "[o]n the one hand, new information may provide further evidence for, and therefore strengthen, old assumptions; or it may provide evidence against, and perhaps lead to the abandonment of, old assumptions" (ibid, p. 109). Ideological effect, however, entails the analysis of the new information at two different levels: knowledge and opinions. This differentiation is based on van Dijk's concept of ideology and his definition of knowledge as the set of "factual beliefs" and opinions as the set of "evaluative beliefs" that, together, form individual and social ideologies (1998, p. 48). Contextual effects, then, are here equated to modification or improvement of the old information, by virtue of the new, at the level of knowledge, whereas ideological effects refer to modification or improvement of the old information at the level of opinions. I would argue that in order for some lasting effect to occur, there needs to be both contextual and ideological effects. However, I would argue that only modifying ideological effects can be attributed to a specific text, whereas improving ideological effects reflect pre-existing ideologies that have been reinforced by the text.

Research design
A set of instruments were employed for the reception study and for the textual analysis. The reception study entailed a qualitative questionnaire followed by a semi-structured interview; the text was approached through a critical multimodal analytical lens. Since the focus is on contextual and ideological effects, the analysis explored how the different actors associated with the Rohingya crisis were portrayed and evaluated both by the viewer and by the documentary. Van Leeuwen (1996) makes a strong case for the close relationship that exists between the representation of social actors and ideology. Although the choice of the documentary and related topic is the participant's, the choice of the social actors analysed is mine but also based on the participant's data. The participant chose the programme and watched it at a time and in a way that suited his schedule: the only restrictions imposed were that the programme had to be of a travel or cultural nature, that it was predominantly in English and that it was not based in the United Kingdom. This research design provided a 'naturalistic' viewing experience whilst foregrounding the participant's role in the media interaction by positioning him as an active media user. This enhanced his engagement with the programme while providing useful information on his ideological positioning through the choice of media outlet/producer. First, the participant completed a qualitative questionnaire around the viewing experience. The questionnaire includes multiple-choice options, open-ended questions and note-taking boxes that explore how the participant interacted with the programme before and after watching it. It is divided into three parts: 1) multiple choice questions that investigate the participant's viewing habits with regard to travel and cultural documentaries. It can be filled out at any point and it serves the function to check whether the research viewing experience reflects the way the participant generally watches these programmes. 2) open-ended questions and a box for notes: this part has to be filled in before watching the programme but after choosing it. This part provides some indications of the motivations behind the choice of the programme, the expectations of the participant and their background knowledge of the places, people and events expected to be found. 3) openended questions and a box for notes to be completed straight after watching the programme. This part records feelings and reactions to the programme and whether or not the latter met the participant's expectations. The pre-and post-viewing notes are then investigated in detail during the interview.
Straight after the participant watched the programme, I did too and then we arranged to meet within one week for the interview. This timescale allowed for the programme to be still 'fresh' in our minds; moreover, it meant that I did not have enough time to carry out a detailed analysis of the text. I think this was invaluable as the interview was also meant to be a process of discovery for me as a viewer: the participant and I approached the interview from a similar level of familiarity with the programme and knowledge of the topics presented in it. This, in turn, helped minimise my influence in steering the discussion towards certain topics, since I did not have an in-depth understanding of the preferred readings and ideologies of the text or of alternative ideological positions and contextual information supporting them. Combined with letting the participant's preand post-viewing notes guide the interview, this research design might have mitigated the effects of the observer paradox in the participant. The interview was also divided in three parts and lasted for approximately one hour. The first part explored in detail the participant's answers to parts 2 and 3 of the questionnaire, focussing on those aspects the participant had deemed to be salient and relevant. The second and third parts of the interview aimed to understand the participant's broader ideology in terms of intercultural and socioeconomic matters. Their analysis, however, is beyond the scope of this paper, which will focus predominantly on the first part.
Finally, I carried out the text analysis. Since documentaries are audio-visual texts, I focused on the modes that are accessible through the auditory and visual sensory channels and, more specifically, on speech, sound, music (auditory), writing and shots (visual). Still images, visuals (e.g. maps), colour, body language and physical appearance (here including clothing/accessories and bodies) were also taken into consideration when relevant to the representation of actors and processes (van Leeuwen 1996) and to matters of connotation (Leech 1983). These last three are the analytical focuses under which meaning-making parameters were considered: treating each parameter in detail is beyond the scope of this paper, but they are all discussed in the social-semiotic approach to multimodality and in the MCDS literature (Kress 2010;Kress and van Leeuwen 1996, 2001Machin 2014;Machin and Mayr 2012;van Leeuwen 1999van Leeuwen , 2005 as well as in cognitive approaches to linguistics and critical discourse analysis in the case of active, passive and stative constructions (e.g. Hart 2014; Langacker 2008). Figure 1 below summarises the sensory channels, semiotic modes and meaning-making parameters analysed.
3 Contextual information about the participant and the text

The participant 3
Max is a male in his mid-30s, born in the UK and raised by white British parents. During his childhood he also lived in an African country for a while and keeps fond memories of those years (Interview, lines 568-571): 568. like we … my dad used to go to church, to various churches to give talks, crazy, 569. like, happy clapping. Churches like Pentecost or churches were people, like, tried to grab 570. my hair, 'cos I was the different person in that crowd, they were all singing and waving. 571. Yeah, loads of like, really like … some joyous moments.
He belongs to my extended social circle and I have known him for about three years. He loves travelling, not only for the cultural experience, but also because he likes the challenge of being out of his comfort zone and to escape the "rat race" that he feels is his working life (Interview, lines 490-506). He predominantly keeps abreast of what is happening in the country and the world through BBC News, the Guardian and the Huffington Post, sometimes prompted by friends on social media (Interview, lines 603-624). Based on this contextual information, I would say Max has a fairly progressive political outlook and an interculturally-sensitive personality.  3 The real name of the participant has been changed to guarantee anonymity.

The text 4
Burma with Simon Reeve (BBC 2018) is a two-part travel series, which is categorised by the BBC as factual/travel under 'genre' and documentary under 'format'. The first episode, which Max watched, is about one hour long and was broadcast for the first time on BBC2 on Sunday, 13th May 2018 at 21:00, and then shown again on five occasions before becoming available on the BBC catch-up service. The peak time slot of its 'premiere' meant that the programme was viewed by 1.47 million people on that occasion alone. 5 The host, Simon Reeve, is a popular television personality and has written and presented many travel programmes for the BBC, including this one.
According to the host, the programme was planned during the summer of 2017 before the events of 25th August 2017 (Reeve 2018). On this date the Myanmar authorities reported attacks to thirty police outposts by men belonging to the insurgent group ARSA (Arakan Rohingya Salvation Army), after which 'clearance operations' by the Myanmar military started that caused the mass exodus of Rohingya Muslims to Bangladesh. Although Reeve claims that they decided to press ahead with the trip despite the deepening of the crisis and not because of it, the synopsis of the programme on the BBC website states that "[i]n this first episode, Simon travels to Burma to find out the roots of this crisisas well as heading to Bangladesh to witness the drama that is still unfolding" (BBC 2018).

Findings and discussion
The findings in this section are presented in the same order in which the data was collected. The first section will analyse the participant's interaction with the programme in terms of the pre-and after-viewing discursive constructions of the social actors involved in the Rohingya crisis. I will then move on to the multimodal analysis of the text and discuss to what extent the participant's interpretation aligns with the programme with regard to the representation of actors and related processes. Next, the ideological effects will be discussed to explore the extent to which the participant's representations may have been influenced by the programme. The actors identified in the documentary as directly involved with the Rohingya crisis are the following six: the Rohingya, the ARSA militants, the Myanmar military, Aung San Suu Kyi, the international community and the MaBaTha Buddhist monks.

The media interaction
Max watched the documentary at home on 6th May 2019, using the catch-up service BBC iPlayer and a television set. I received a scanned copy of the completed questionnaire on 7th May and the follow-up interview took place on 9th May at his house, after I had watched the programme myself. The discussion in this section is based on both the questionnaire (henceforth Q) and interview (henceforth I) data. 6 The participant data suggest a number of reasons behind the choice of the programme: An interest in travelling and in Burma as a destination, some prior knowledge of Burma's history through schooling, a liking of Simon Reeve and his programmes and an interest in finding out more about Burma's current affairs. One of Max's answers in part 2 of the questionnaire states the motivation for choosing this documentary: "I enjoy watching travel programmes and I have always been interested in Burma and the fact [it is] so closed off from the rest of the world." (Q, item 9). The interview gave Max the opportunity to expand on this (I, lines 24-40): 24. So, through travel I was quite interested in Myanmar, or Burma, 4 Unless otherwise stated, all the information about the programme can be found at https://www.bbc.co.uk/programmes/ b0b3g1nm (Accessed on: 28 April 2020) 5 Figures available at https://www.barb.co.uk/ (Accessed on: 2 March 2020) 6 In the interview data M stands for Max and R for researcher. From the above quotes it is clear that Max did not choose the documentary randomly, but on the basis that the information he could gather about it fitted his ideological milieu. Moreover, the motivations expressed seem to confirm he chose the programme because it interested him and not because he thought I would find it interesting. Finally, Max's answer hints at his understanding of the programme as a means through which he can gain insights on a part of the world and a culture he has not been able to access through direct experience. This could be interpreted as a sign of trust in both the host (Simon Reeve) and the producers (the BBC) as agents that have permission to influence his knowledge and opinions, and therefore as granting the text a licence to have some ideological effects on him.
Answers to the next two items in the questionnaire address what Max knew already about the programme, e.g. through reading a synopsis or watching a trailer (item 10), and what places and people he expected to find in it as well as ideas he associated with them (item 11). As for the first point, Max wrote: "I am familiar with Simon Reeve and have watched his programmes visiting other parts of the world. I have spoken with friends about the episode and know it touches up the recent Rohingya refugee crisis" (Q, item 10). As for the second point, the box below contains his answers (Q, item 11): Again To summarise, Max was already aware of the Rohingya crisis as something that was currently going on in Myanmar and also knew the programme would address it, but not to the extent that it actually did. From his notes, it is also clear Max was aware of some of the history of Myanmar, here including the British colonial past, some form of oppressive government, its relative isolation from the rest of the world as well as expectations to see things would be different now borders are open, and the fact that Aung San Suu Kyi has some kind of role in the country's history. Once again, this information confirms that the text was deemed to be 'ideologically compatible' and that Max chose to interact with it thinking he would gain some benefits from such interaction beyond mere entertainment. The answers in part 3 of the questionnaire focused on what Max thought of the programme after watching it and what aspects had impacted him the most. The note-taking box with Max's thoughts is reproduced below (Q, item 12): The next answer is also reported in full (Q, item 14): With regard to the actors portrayed to be involved with the Rohingya crisis, as well as confirming and expanding on the people noted down on the questionnaire, during the interview Max talked (unprompted) about the two members of ARSA. A summary of Max's representation of the different actors before and after viewing the programme is provided in Table 1 together with the analysis of potential contextual and ideological effects. The summary takes into consideration information provided before watching the programme and information provided after watching the programme that refers to Max's ideological state prior to watching it. The table is best read vertically by column. The findings from the media interaction serve two purposes. Firstly, they allow us to check whether Max interpreted the social actors in the programme through a dominant, oppositional or negotiated code: the results of the multimodal analysis will provide answers to this matter. Moreover, they indicate the contextual and ideological effects that might have taken place as a consequence of watching the programme.

Multimodal analysis of the social actors involved with the Rohingya crisis
The analysis takes into consideration how the modes of speech and writing (labelled language), shots (labelled visuals), sound and music operate individually in representing actors and related processes and how their co-Sympathy to the plight of Rohingya. Shocked at the scale of the issue and how it is not a recent issue (been happening for decades). Surprised by existence of "militant" Buddhist monks. Disappointed that liberation/progress doesn't appear to have happened beyond the city. Amazed by existence of Bagan.  occurrence offers a specific representation, hence encouraging a preferred reading of the text. The programme can be divided into eight parts based on different thematic focuses and signalled by breaks (e.g. fade-to-black shots). Figure 2 below shows the distribution of the social actors across the programme:

Exploring media effects: MCDS and Audience Research
As Figure 2 shows, some actors only appear in specific parts, e.g. the MaBaTha monks on whom Part 5 focuses, while others, e.g. the Military, are referred to throughout the programme, although not necessarily in connection with the Rohingya crisis. Figure 3 offers the complete analysis of the MaBaTha monks as a sample: Finally, Table 2 provides a summary of how the modes are employed in the representation of the social actors together with my interpretation. Moreover, I have specified whether or not the actors are allowed a voice that is the possibility to express their opinions. iii. monks shown as active, doing things; iv. The senior militant monks are portrayed in a static, authoritative position of power (low camera angle); they are the "new" information, shown at an intimidating close distance (as they are shown as very grave or as laughing at some of the serious concerns put forward by SR, e.g. whether they see Islam as a threat); v. Students and teachers are initially shown from a "safe" distance and then brought closer to the audience while they are watching the video; the teachers and students are then taken to a closer distance while speaking, which can be seen as intimidating by the audience as what they are saying is either very graphic and depicting horror or making generalised accusations such as "most Muslims are intrinsically mean", to which a shot of a laughing (happy of the results of their preaching?) monk follows. Exploring media effects: MCDS and Audience Research Table : Summary of the multimodal representation of the social actors involved in the Rohingya crisis.

Language Visuals Music/Sound Overall Interpretation
The Rohingya Voice: possibility for the Rohingya to tell their side of their story.
Close-ups to create an emotional connection with the viewer; wide and "busy" shots to highlight scale and nature of their present condition.
Either grave music or no music at all, especially during interviews to eliminate distractions in the soundscape. Often "live" sounds and noises to highlight authenticity.
The Rohingya are represented as the victims both through the host and their own representation. The semiotic modes operate to create an emotional, sympathetic connection with the viewer.
Connotation is generally positive and compassionate.
Active processes are mainly positive, portraying the Rohingya as struggling to save their lives. The negative ones (with the exceptions of one instance) are spoken by the MaBaTha monks.
Passive processes and statives are largely positive in the sense that they portray the Rohingya as victims.
When the statives are negative, they are spoken by the MaBaTha monks.

The ARSA Militants
Voice: possibility for the jihadists to tell their side of their story.
Faces not shown or pixelated to guarantee anonymity, a sign of respect.
Grave, slow and melancholic music. Also sound of rain heightened.
The militants are represented as victims who fight back because of exasperation. The semiotic modes operate to create an emotional, sympathetic connection with the viewer.
Connotation is generally sympathetic (resistance, forced to, inspired to). Active processes are generally negative when spoken by both Simon Reeve and the militants. Passive processes and statives, however, are mainly positive, in the sense that they portray the militants as victims.

The Military
Voice: no possibility for the military to tell their side of their story. If their views are conveyed it is through Simon Reeve.
Hardly ever shown. In the couple of shots where they are represented, they are uniformed and armed.
Grave melancholic music when describing the actions against the Rohingya; cheerful music (parts  and ) when talking about some "oddities" of the generals; suspense-type of music during the "action" shots with the wounded Rohingya. Also sound of rain heightened.
The Myanmar military is represented as the culprit and bears the responsibility for the current situation. They are, however, an invisible, faceless and nameless threat. The semiotic modes portray them as cruel and irrational at the same time; a political player that cannot be trusted.
Connotation is generally negative, not so much in the names used for the military, but for the words used to describe their actions. Active and passive processes as well as statives are overwhelmingly negative.

Aung San Suu Kyi
Voice: no possibility for ASSK or the government to tell their side of the story. If their views are conveyed is through Simon Reeve.
ASSK is only shown twice through a newspaper photograph.
Gentle Burmesesounding music in initial part; grave and melancholic music in the final parts. No music at times. ASSK is represented as a travesty and accomplice to the military by way of not denouncing their actions. She is depicted as traitor of

Discussion
From a comparison between Max's evaluation of the actors and the multimodal analysis, it appears that Max has interpreted the representation through a dominant code that is he has correctly identified the preferred reading of the programme and has agreed to it. With regard to the Rohingya and to the ARSA militants he has empathised with their plight and justified their violent actions. As for the military, he has not questioned the negative representation created by the programme, and the same goes for the current Myanmar political leader, Aung San Suu Kyi. He has not challenged the positive representation of international players, whether in their financial and expert roles (investors and the United Nations) or in their humanitarian role (UNICEF). Hardly ever shown. The UNICEF centre and members of staff are the only exception. A very humanitarian face.
Grave or melancholic music.
The international community is portrayed as neutral, expert and humanitarian. The semiotic modes operate to instil trust in them as the only political player that is already doing charitable actions to alleviate the sufferings and that would guarantee freedom and progress.
Connotation is neutral or positive when talking about political institutions, but negative when discussing their assessment of ASSK; positive when talking about UNICEF and 'investments'. Active and passive processes as well as statives follow a similar pattern to connotation.

MaBaTha Monks
Voice: possibility for the monks to tell their side of their story.
Close-ups to create a sense of threat for the viewer. Often low camera angle to depict them as powerful and influential.
Traditional Buddhist music; grave solemn music; no music during interviews to eliminate distractions in the soundscape.
The militant monks are portrayed as part of the ideological justification for the suffering of the Rohingya. They are shown spreading propaganda and indoctrinating younger monks to stir them away from the true, compassionate essence of Buddhism.
Connotation is generally negative when Simon Reeve is speaking and positive when the monks talk about themselves. Active and passive processes as well as statives follow a similar pattern to connotation, i.e. dependent on who the speaker is.

Exploring media effects: MCDS and Audience Research
Finally, he seemed in agreement with the negative representation of the MaBaTha monks as spreading hatred against the Rohingya and providing the ideological justification for the military and political response to the crisis. Figure 4 below provides a summary of both Max and the programme's representations for reference: However, this preliminary comparison, much along the lines of the recontextualization studies, provides only some understanding of the complex interaction between Max and the programme. One way to explore the extent to which the programme has had some effects on Max's ideological milieu is through the concepts of contextual and ideological effects.
Two actors jump out of the comparison between the text and Max's representations: the military and the international community do not seem to have had significant relevance in Max's interaction with the text. In the case of the military, although they are frequently represented in the programme, their representations did not improve Max's knowledge nor modified his negative ideological position toward them as being an oppressive entity. Likewise, although the international community appears often in the programme, either in the form of faceless 'investments' or embodied by international aid organizations such as UNICEF, their representations did not add new information or modify Max's positive opinion of their beneficial impact from an economic and humanitarian point of view. Arguably, the only extra information gained in this regard is that the benefits of the open borders stopped in the main citiesan issue, however, that Max blames on the Burmese government for failing to facilitate the process by not building infrastructures rather than to international dynamics. Moreover, the fact that specific companies behind 'international investments' are not mentioned in the text is an exclusion (van Leeuwen 1996) that could serve the ideological purpose of preventing viewers from evaluating such investors, while at the same time minimising contextual and ideological effects by increasing the cognitive effort required to identify the investors.
Secondly, there seem to be some contextual and improving ideological effects occurring for two social actors. For the Rohingya people, the contextual effect is an improved knowledge of the conditions in which  they live, namely how long the relationship between the Rohingya and the authorities has been problematic, the scale of displacement and the size of the refugee camp. There didn't seem to be any modifying ideological effects, but rather self-deprecation for not being aware of the scale of the problem earlier. The multimodal representation of the Rohingya as victims reinforces Max's sympathetic stance toward their cause, hence resulting in improving ideological effects. A similar pattern can be observed with the ARSA militants, the newly acquired knowledge of whom represents the contextual effect. Again, however, Max seems to have already held the opinion that violence can be justified under certain circumstances, specifically for being the victims of abuse and dramatic sufferings. As for the Rohingya in general, the sympathetic representation of the militants in the programme strengthens such ideological position. Finally, there seems to be significant contextual and modifying ideological effects occurring with Aung San Suu Kyi and the MaBaTha monks. Regarding the political leader, the contextual effect consists of novel information regarding her role in the crisis and, possibly, her role in the failure to spread the benefits of open borders throughout the whole country. This in turn seems to produce the modifying ideological effect to challenge Max's previously held assumption of people being "more liberated" since the democratically elected government of Aung San Suu Kyi has been in power and the added disappointment of her not standing up to the military and not condemning the violence in Rakhine State. Once again, the multimodal representation encourages such an evaluation of this social actor. Finally, the same pattern can be seen for the MaBaTha monks: the contextual effect is provided by learning about the existence of such a religious faction. The modifying ideological effect can be identified in the reconsideration of Buddhism as a peaceful faith and the accommodation within the existing ideological milieu of the possibility of an extreme component of such religion. Yet again, the multimodal representation of the programme supports this.
To summarise, Max's representation seems to result from interacting with the text through a dominant code that aligns with its preferred reading. This does not necessarily mean that Max was influenced by the text with regard to all the actors, since he may have already held some of those opinions. Indeed, this seems to be the case for the military and the international community. Likewise, although there are some contextual effects in the case of the Rohingya and the ARSA militants, Max seemed to already hold the same ideological positions presented in the text, which nonetheless strengthened his own through improving ideological effects. As for the MaBaTha monks and Aung San Suu Kyi, there seem to be some modifying ideological effect represented by newly formed opinions about Buddhism not being an entirely peaceful religion and about the Myanmar leader not being the catalyst for change that the international community thought she was. A research hypothesis that could be tested with a second qualitative investigation after a significant lapse of time is whether the modifying ideological effects have indeed entered Max's ideological milieu.

Conclusion
The integration of AR provided some useful insights. First, it can be argued that the analytical concepts of Kress and van Leeuwen's social-semiotic approach, largely based on principles derived from Halliday's Systemic Functional Linguistics (1978), and integrated by some insights from cognitive linguistics (e.g. Hart 2014; Langacker 2008), can be fruitfully utilised for the analysis of documentaries, at least as far as the representation of social actors is concerned. The contextual and ideological effects analysed are compatible with the preferred reading suggested by the multimodal analysis. This point should be taken into consideration as part of the on-going discussions regarding the nature and principles of the growing field of multimodality (Bateman 2018;Ledin and Machin 2018b;O'Halloran et al. 2018).
Furthermore, the notions of contextual and ideological effects operationalised through the AR methodology gave us some indications with regard to a) what aspects of a text the viewer found more relevant vis-à-vis their existing ideologies; b) what aspects of the viewer's ideology pre-existed the media interaction (signalled by improving ideological effects); and c) which ones resulted from it (signalled by modifying ideological effects). However, one qualitative investigation cannot provide a definitive analysis of whether aspects of a text result in stable changes in the viewers' ideological milieu. One way to take this approach forward can therefore be Exploring media effects: MCDS and Audience Research qualitative longitudinal studies that aim to 'track' the ideological development of viewers and explore further the possibility of modifying ideological effects through multimodal communication. Moreover, it is necessary to see whether similar processes hold for different interactions between individuals and programmes. Finally, ideologies that are not picked up by a viewer should also be analysed, as this can direct us towards those ideologies that are so naturalised as to be hegemonical and suggest a focus for counter-hegemonic discourses.
The integration of a full-blown AR methodology can enable the critical researcher to consider those aspects that the reception methods employed in (M)CDS so far could not, i.e. motivation and ideological reasons for interacting with a text, an understanding of someone's ideological negotiations with a media text and the effects texts may have on an audience. Reception studies within multimodal research suggests that each individual's interaction with a multimodal text is different and AR can assist us in reaching a nuanced understanding of the cognitive processes at play. However, AR, like most other qualitative approaches is still open to issues such as bias and the observer paradox. A careful research design is necessary to minimise these and the one proposed here suggests ways in which this can be done: i) allow the participant to freely select the text; ii) carry out the follow-up interview from a viewer (rather than analyst) position; iii) differentiate between prompted and unprompted data and basing the analysis primarily on the latter. Ultimately, it is suggested that integrating AR in MCDS is a fruitful way to investigate the thorny issue of media effects and the cognitive mechanisms that underpin this phenomenon.