Slumming on Social Media? E-Mediated Tourist Gaze and Social Representations of Indian, South African, and Brazilian Slum Tourism Destinations

: Slum tourism is a hotly debated genre of travel. While it may foster intercultural encounters with marginalised “others”, it is also accused of reinforcing stereotypes and exploitation. Both aspects are ampliﬁed by the communication through social media of the slum tourism experience, that contribute to challenge or conﬁrm stigmatizing representations of slums and their inhabitants. Based on the theoretical constructs of the tourist gaze and of social representations, this article addresses this particular type of digital contact. A lexicometric approach was used to analyse an extensive corpus of reviews on TripAdvisor ( N = 8126). The ﬁndings not only conﬁrm common themes already identiﬁed by the literature: the eye-opening component of touring poverty and the gatekeeping function of guides; but also show the emergence of context-dependent speciﬁcities, such as a hedonistic feature in the Cape Town region; or the integration of favelas within the representations of the city of Rio de Janeiro. Furthermore, the results show the tension between the “othering” and the “sameing” mechanisms, making this tourism practice a space in which shallow and deep tourist gazes interact and co-exist, and are crucially mediated by the gatekeeper of the tours: the guide.


Introduction
Slum tourism is a controversial and fast-growing type of tourism in which tourists spend some time, from a couple of hours to a few days, visiting, touring, and experiencing the way of living of locals in shantytowns. Slum tourism has two faces and its ethical premises have been often questioned [1]. On the one hand, this type of tourism is deemed voyeurism by wealthy tourists consuming and exploiting the slum with the aim of touring poverty [2]; furthermore, the zoo metaphor is often used by critiques to stress the similarities between tours and safaris and the dehumanization of slum dwellers [3]. On the other hand, the "witnessing" dimension promoted by this phenomenon allows tourists to enter into contact with the "other", to engage with issues of inequality and injustice and, therefore, to possibly be the trigger for social action and social change [2,4].
Social media have a significant role to play in this scenario, not only because local tourism operators in slums mainly use online channels for marketing their businesses [5], but also because they are a privileged space for intergroup contact [6]. Indeed, the low barriers to entry, the pervasiveness, and the reduction of physical and affective barriers allow individuals to anticipate, express, and co-construct their representation of the others in a potentially supportive context. Yet, at the same time, normative and communicative phenomena, such as reduced accountability and eco chambers, could reinforce intergroup hostility and prejudice [6][7][8]. Nevertheless, the importance of social media in slum tourism has been examined thus far only to a limited extent [2,9].
Based on the theoretical construct of tourist gaze [10] and on social representation theory [11], this research provides the most extensive analysis of tourists' experience in Brazilian, Indian, and South African slums, based on their online reviews (namely seven years of TripAdvisor reviews). The current study aims at contributing to the debate on slum tourism as a peculiar form of intercultural experience, and at understanding the role of tourists as sources of extended contact. These will ultimately open the floor for a discussion about the power of social media to co-create the tourist gaze. Additionally, by illustrating a method of statistical lexical analysis suitable for examining large textual corpora, the article aims at contributing to the analysis of intercultural experience, illustrating a way to generate a synthetic understanding of the way intercultural encounters are discussed on social media.

Slum Tourism and Its Controversies
Slum tourism is a fast-growing, global phenomenon, involving more than 1 million tourists every year [1]. Its origins are in Victorian London and early twentieth-century USA, where rich people used to leave their better-off neighbourhoods to go and see how poor people lived in the underprivileged parts of the city [12,13]. In its contemporary forms, slum tourism started in the 1990s in South African townships and Brazilian favelas and spread all over the developing world, with India leading the way in the Asian continent [14].
A typical slum tourism visit consists of a 2-3 h tour, often including some live music or dance performances as well as an exhibition of local craft products and their production system [14]. Socially relevant activities, such as a local nursery or day-care centre, can also be part of the visits. Besides this classic structure, the touristic experience can also include a culinary dimension with local restaurants and bars offering typical dishes or drinks, as well as the possibility of finding accommodation and sleeping in the slum [5]. Despite the potential number of new activities and jobs, socio-economic regeneration and pro-poor development promoted by slum tourism are a matter of debate. Local ownership of tourism products, political and financial support, and locals' participation in decision-making are critical factors for developing responsible tourism in these suburbs [15][16][17][18].
In the slums where tourism is present, the phenomenon not only affects the economic and social structures of the communities but also how they are pictured and perceived [19]. Tourism and tourists indeed contribute to shape the representation of these destinations and of the overall representation of what slums are. However, as already noticed, the role of slum-tourism is also controversial in this regard. On the one hand, slum tours may promote positive experiences of contact with locals, which can modify tourists' attitudes, promote intercultural understanding, break the isolation of slum dwellers and challenge the stereotypes that threaten their identity [5]. On the other hand, stereotypes can be reinforced, sometimes even by locals, in the attempt to meet the quest for "authenticity" expressed by tourists, which is their intention to experience the local culture and the actual way of living in the slum [12,13,20].
Frenzel and colleagues [1] suggested that visitors give values to the slums' inhabitants using two alternative processes: "Othering", which places value in someone who is different and is thus a source of new insights and "Sameing", which assumes a similarity among visitors and locals in fighting against inequalities. However, other authors give the opposite value to the two processes. Othering can be used to activate social comparison in terms of absence (i.e., not modern, not wealthy) and to confirm the negative stereotype of the slum and its inhabitants, providing a new version of the bon savage. Sameing can be used to pretend erasing differences and to make them invisible to the eyes of the dominating group, thereby maintaining structural social inequalities [21]. The psychosocial processes embedded in the slum tourism phenomenon therefore call for a reflection on the way representations are created, circulated, and used by the actors involved in it (e.g., guides, dwellers, tourists, businessmen, film-makers, politicians, etc.), and on how this typology of intercultural encounter constructs what a slum is and who its dwellers are [22,23].
In order to generate a better understanding of the phenomenon, this article builds upon the concepts of tourist gaze [24] and of social representations [25].

Social Media Tourists, Gazing at the Destination
The tourist gaze [24] can be defined as the set of expectations that tourists place on local populations/places in the search for an 'authentic' experience. In his seminal work, Urry [10] proposed a framework in which the tourist gaze is elaborated as a particular way to see the world, primarily conditioned by the imaginary and/or the representation created in the first instance by the destination [26]. Other authors suggest how, in a similar way, residents look at tourists, projecting expectations and coercion onto them and their behaviour [27][28][29]. The rise of online social networks somehow challenges these concepts and, consequently, the touristic experience [8].
Today, the tourist gaze (but the same could be argued for the host gaze) happens in 'mediascape' where tourists can violate existing imaginaries promoted by the destination and can contribute to creating new ones [30]. Tourists not only 'post' 'travel tales' to significant others but, thanks to dedicated social media platforms (e.g., Tripadvisor), address entire travel communities of prosumers who engage in self-reflection on their experience, while reading others' accounts, in a veritable process of 'intertextuality' [31].
This was discussed by Robinson [32], who named the phenomenon "e-mediated gaze": the possibility offered by websites to perform instant and spatially positioned activity of recording images, to build informal pictures of a destination, and to narrate countless stories of a place over time [32] (p. 358). Over the years, the role of digital photos hosted online gained traction in the academic world [33] with some authors referring to this phenomenon as a new type of gaze mediated by specific social media [34]. Additionally, theoretical and empirical work by Cohen [35], McKercher [36], and Plog [37] support the perspective of different gazes occurring from different stance locations (psychological, sociological, and behavioural outlooks). Regarding psychological differences, Plog [37], for example, discusses various types of gaze(s) based on traveller personalities, presenting a continuum ranging from 'Psychocentric' travellers, who seek safe experiences and engage in a shallow gaze, to 'Allocentric' ones, who seek novel and authentic experiences and engage in a deep tourist gaze. Woodside and Martin [38], who elaborated the Tourist Gaze 4.0, discuss the 'psychological indexing process' by which travellers automatically anchor their experience and representation of a destination with their own life experiences stored in memory.
Research in this field is however limited, as studies mainly focused on pictures while extensive analysis of online tourist reviews are lacking [33,39,40]. Similarly, some authors have explored locals' perceptions of tourists, identifying the potential for both of this liminal experience [23,41,42]. Furthermore, within the specific literature on slum tourism, despite the relevance that social media play [43,44], few scholars have investigated how the internet and social media are contributing to shaping narratives about intercultural encounters. Studies mainly involve in-depth qualitative analyses conducted on a limited amount of tourist reports [2,9,45].
Therefore, it is argued here that further extensive analyses of the nexus between slum tourism and social media are needed. The term 'social media gaze' [46] will be used in this work to refer to the contents posted on social media and created by internet users which contribute to construct and challenge shared representations of slums and their inhabitants.

Constructing the 'Others': Social Representation Processes
A second point of reference for the current study is social representations theory (SRT) [25]. According to Moscovici, social representations are "systems of values, ideas, and practices, which have a twofold function. First, they aim to define an order to shape the material and social world and control it; second, they aim to make it possible to commu-nicate among the members of a social group or community by supplying them with a code for naming and classifying the various aspects of their world and their narrativity." [47] (p. XIII). SRs can be conceived as new forms of common sense, which are co-constructed in everyday communication with other members of the groups we belong to, in order to give meaning and to cope with relevant objects. Two social processes are at the core of SRs: anchoring and objectification, "Anchoring involves the naming and classifying of new encounters, ideas, things or persons. It is based on an existing order of meaningful names. Objectification solidifies and makes tangible the abstract and potentially threatening new idea." [48] (p. 172). The social dimension of SRs is particularly underlined by the sociodynamic approach. Briefly, this perspective considers SRs as "organizing principles of symbolic relationships between individuals and groups" [49] (p. 2): once shared meanings around a debated object are defined and structured, people take systematic positions in this representational field which reflect psychological characteristics, sociological belonging, and psychosocial processes including intergroup relationships and knowledge [50].
The importance of imagery processes is finally key to SRT. In fact, "whether an image is accepted or not by a group is neither a problem of truth nor an arbitrary choice, but determined by the group's experiential world and the negotiated consensus of the group members" [51] (p. 10). Images and the processes of selecting, framing, and creating images have been treated as a doorway in the study of SRs, as products, source, and forms of transmission of SRs [52,53].
These overall processes are relevant to our understanding of intercultural encounters and mediated experiences of slum-tourism, especially if we consider that in-groups and out-groups can be objects of representations as well [54]. The shared knowledge of 'who we are' and 'who they are' [55] serves to maintain or contest inequalities, to protect a positive image of the in-group, to empower communities, and orient political action [56,57]. SRT suggests that inclusion or exclusion of 'the others' are achieved through the process called 'placing into otherness' (mise en altérité); that is, by representing 'the others' as carriers of cultural values and/or judicial characteristics different from the ones shared by in-group members [58][59][60]. Strictly connected with processes leading to prejudice and marginalization, SRs of the 'others' as different lead to the construction of veritable "stigma theories", which emphasize the inferiority of the out-group and justify the social exclusion of marginalized groups [61] (p. 232). An example of processes that serve to maintain the status quo is the tendency to represent dominant groups as a collection of individuals, whereas subordinated groups are represented in terms of aggregate; that is, of homogenised, depersonalised, interchangeable others [62]. Marginalised groups can, of course, resist such stigmatising SRs [54]; however, different actors exert different power in the process of constructing, imposing, or resisting the representations of self and others' identities. In this regard, social media have been suggested as an ideal arena where representations of the self and of the 'others' can be challenged, resisted, and contested especially when other public spaces are not viable [63]. Nevertherless, from a SRT perspective, there is a dearth of research on the role that social media increasingly plays in social construction processes [64,65].
Moreover, the SRT has also been marginally used in tourism studies. SRT has been adopted as a framework especially with regard to competing views of tourism, of tourists, and their impact on local communities [66][67][68][69]. There is evidence of the use of social representation theory to analyse attitudes towards, and responses to, tourism development [70][71][72] as well as to analyse collaboration among stakeholders involved in tourism planning and activities [73]. SRT has also been adopted to explore mutual gaze on practices such as nudism or suntanning [74], or as a way to connect tourism activities with other societal issues such as climate change [75]. Looking more specifically at the social construction of the tourist destination [76], the few examples that connect SRT with tourism, identity, and online communication mostly take into consideration official destination websites [77,78].
The analysis of tourist reports is thus pivotal to examine the coexistence of different SRs of destinations and of their inhabitants, and the connected intergroup and intercultural processes (e.g., to define groups identities and boundaries, to legitimize one's own experience, to evaluate the 'others').

Aims
Based on the analysed literature, this research has been designed to shed light on three major issues at the intersection of the tourist gaze, social representation, and social media in the field of slum tourism. These are: - What are the shared meanings attached to different slum tourism destinations?
The research elaborates on expectations tourists place on local populations. Drawing on SRT, the study will explore similarities and differences in the social representations of three main global destinations: India, South Africa, and Brazil. These social representations form the social media gaze of the destination and, at the same time, serve as a mediated form of contact that contribute to future tourists' experiences. - How is the slum tourism experience mediated by the 'others' in interactional contexts?
This study has been designed to shed light on the social construction of the tourist gaze. In particular, the study discusses the importance of the actual interaction with mediators; this is what Larsen, Urry, and Axhausen [79] describe as corporeal proximity and Castro [80] as the interaction context (i.e., who is with the traveller and who, if anyone, is mediating the experience).
-How do social media gaze and social representation processes contribute to the digital mediation of intercultural processes involved in slum tourism?
The study explores the 'othering' or 'sameing' process [12,59], with respect to the slum experience and discusses how they contribute to positive and negative impacts of this type of tourism on local population and intergroup relationships.

Site Selection and Data Collection
The study analyses and discusses the above-mentioned research questions mining the text of 8126 online tourist reviews on TripAdvisor for three popular slum tourism destinations: India, South Africa, and Brazil.
Brazil, South Africa, and India are well-established and documented destinations for slum tourism, a practice presenting differences both among countries and among slums of the same country [12,81]. In order to identify the most relevant destination to be examined, the researchers selected the ten most popular touristic cities in each country, and the biggest 20 slums for each city, retrieved from each country's official data. To this corpus, slums mentioned in the tourism literature were also added. The result was a list of shantytowns and their respective cities. Then, the researchers performed a manual search in TripAdvisor to identify which slum returned relevant results and only the ones with relevant data (Clusters and Reviews) in the social media platform were included in the final list. This result was finally used on software designed ad-hoc to retrieve data using the name of the slum as the search word and the city as the location. In addition to the actual names of the slums, we noticed that many TripAdvisor contributors used general terms (e.g., slum) in their reviews to post details of their slum tourism experiences: the most commonly used terms were 'slum in India', 'township in South Africa', and 'favela in Brazil'. Recent literature debates and challenges the definition of slum tourism in historically rich locations, such as Soweto in South Africa or intertwined and interconnected to city neighbourhoods, such as favelas in Rio de Janeiro. It is argued that this type of tourism should be defined as cultural tourism or heritage tourism [42,82,83]; however, we decided to include the terms in our search, as this helped to considerably broaden the corpora. For the same reason, we decided to add 'township' to the searches for South Africa; although the term is not a perfect synonym of the term 'slum', slum tourism literature also addresses and encompasses tourism taking place in townships [84] and in the conceptualization of TripAdvisor reviewers, the two terms are blurred. Including these general terms in our search allowed us to enlarge the corpora by 572 reviews for South Africa, 1538 reviews for Brazil, and 1173 reviews for India.
Only reviews initially written in English were included in the sample; we report in Table 1 the details of the reviews and illustrative variables retained after manual screening of the data.

Data Analysis
The corpus was investigated through lexicometric analyses, supported by IRaMuTeQ software [85], using techniques that are particularly suitable when dealing with massive digitised data [86]. This approach differs from other automatic or semi-automatic content analysis often applied to web sources: it does not count the frequency of pre-defined key terms (e.g., like in so called sentiment analysis), neither does it use concept thesaurus to automatically classify contents (e.g., like in topic modelling). It rather follows a bottomup approach: contingency tables (e.g., words by countries) are submitted to a variety of statistical analyses aimed at identifying statistically significant vocabulary in a given set or subset of data, which are then interpreted as anchoring categories and ways of thinking about a given object [87]. In order to perform the analyses, as usual, the corpus was preprocessed to reduce data dispersion and ambiguities. Repeated segments were reduced to a single form (e.g., SA and South Africa into south_africa). Finally, the text was lemmatised using the software internal dictionary for English (Table 2).
A first output is the list of the most frequent forms, which provides a first idea of the shared contents used in the description of slum tourism experiences across the countries interpreted.
Secondly, a specificity analysis was applied to compare lexical forms used in accounts related to Brazil, South Africa, and India. Specificities analysis is a test based on hypergeometric law that reveals which terms are significantly (p < 0.05) over-used (or under-used) in a part of the corpus compared with the overall distribution. In our case, the specificity analysis was applied to compare lexicons used in describing the experiences in South Africa, Brazil, and India.
Lastly, each of the three sub-corpora (South Africa, Brazil, and India) was submitted to descending hierarchical classification (Reinert method, [88,89]. This method is based on three main automatic steps: first, each text is fragmented in context units, which approximate meaningful segments in a discourse (e.g., a sentence identified by punctuation); second, a correspondence factor analysis is performed on the matrix context units by forms; third, an iterative algorithm is applied to identify partitions that maximise the inertia inter-classes. The result is a clusterisation of the corpus that summarises the main themes discussed in a given corpus. Each cluster is interpreted on the basis of the forms that significantly contribute to it (chi square statistics) and the illustrative variables that are significantly (p < 0.05) associated to those contents. Typical context units and text can then be identified by computing the sum of the chi-square values of the marked forms in each segment. It is interesting to stress the high frequency of comments related to safety: Safe (1409).

Descending Hierarchical Classification: Typologies of Discourse
Two common themes emerge from the analysis applied to the South Africa, Brazil, and India sub-corpora (Table 3): the importance of tour guides and the eye-opening effect that this intercultural experience has on tourists. Note to the table: Clusters are extracted from each country's sub-corpus, percentage refers to classified segments included in each cluster, the most relevant terms that contribute to defining each cluster are reported.
The practicalities of the tour were discussed both in India and in Brazil, whereas emphasis on the touring of community projects was shared by reviews related to South Africa and India. The analysis then shows the peculiarities of each context: a focus on historical sites and on local culture (food) emerged in South Africa, whilst the Brazilian corpora has a theme related to sleeping in the favela. In the following section, we discuss the emerged clusters and themes (Table 3).

Shared Themes among Destinations
As mentioned above, the role of the tour guide as the key mediator of the slum tourism experience is present in all three corpora.
In South Africa, the guides are mainly connected with organisational aspects of the tours (pick, hotel, bus) and with a particular focus on the Felleng guided tour in Johannesburg. Here is an extract of text emblematic for this cluster: "We spent a full day with Agnes in Pretoria and Soweto; in Johannesburg it was truly a life changing experience; we are now supporting Kyp as a way to share our appreciation for Africa; Agnes was delightful, knowledgeable and flexible; we recommend Felleng tours".
In Brazil, this cluster is characterised by the given names of guides, such as Patrick, Thiago, and Pedro, and of specific tours, such as the Favela Tour, and appreciation for the knowledge and personal characteristics of the guides. "An awesome tour round a Rio favela which gives a great insight into life in the favelas. Patrick, the tour guide, is really knowledgeable and ensures a safe and informative tour; highly recommend this tour when visiting Rio". These contents are significantly present in reviews of places generically labelled 'Favela' and not linked with any specific location.
Finally, this cluster in India is related to terms describing the guides' characteristics in terms of knowledge and of attitudes with the clients; guides are represented in positive and enthusiastic terms by the reviewers. "Our guide, Chetan, was extremely knowledgeable; there was no question he couldn't answer. He was also really friendly and helpful, spoke excellent English and just generally looked after us so well". Reviewers associated to this cluster are young tourists (aged 18-24 years) and the experience focus is mainly on the slum of Dharavi.
The second cluster present in all three corpora relates to the eye-opening experience connected with slum tourism, an intercultural encounter often discussed in the literature.
In South Africa, this cluster collects the feelings of visitors when touring slums, especially during slow tourism (e.g., bike tours) and visiting non-profit initiatives. The eye-opening feature, so often reported by the literature, for this type of tourist and the reflection on safety and poverty typical slum tourism practices are narrated here: "saw a range of the types of houses that people live in; the bikes were well maintained. I felt safe and looked after during the ride; the atmosphere in the township felt very positive although there were some shocking living conditions and very poor people". Variables associated with this cluster suggest that this experience is linked with Khayelitsha and Masiphumele and less known slums (Alexandra Township in Johannesburg, and other unspecified townships) and the authors' profiles suggest that these are aged 25 to 49 years. In the Brazilian case, this cluster shows a traditional view of slum tourism, reporting the eye-opening experience and feelings by narrowing people living in poverty. A fascinating element of this cluster, which deviates from the 'othering' dangers reported by the literature, is the use of both collective terms, such as 'community', and of terms referring to individuals: 'person' and 'resident'. This text segment exemplifies reviews present in this cluster: "Olivia knows everyone in her favela so we always felt safe under her wing; we also gained remarkable insight into the lives of the people living in the favelas and an overview of the social initiatives underway to improve their living conditions". The reviewers belonging to this cluster are from North and South America and relatively experienced (age 35-49, 50-64, and 65+ years). The eye-opening experience is significantly prevalent in reviews related with Rocinha (Brazil). In India, this dimension of slum tourism is particularly present, as two clusters describe the touring experience as eye-opening and memorable. A typical text segment related to the first of the two clusters is: "no matter what tour you choose with reality, no matter which guide you have, you are guaranteed to have an incredible experience in Mumbai; I highly recommend them to anyone and everyone".
Again, authors in this cluster are young tourists, aged 18-24 years; the experience reported by the online reviewers is compared with the imaginary they had before the travel and shaped by the well-known movie Slumdog Millionaire. This is one of the recurrent discourses in our comparative analysis and it is in line with previous literature. Reported below is an example of a text segment belonging to the second of these clusters: "we took the slum tour and really did not know what to expect. It really was a fascinating morning and a real eye-opener, it was amazing to see how enterprising and hard-working the people are and very humbling to really see the conditions they live in". Not surprisingly, the slum of Dharavi is the one related to this cluster and, once again, reviewers contributing to it are young tourists, aged 18-24 years.
Brazil and India shared another cluster related to the organizational side of the slum tour experienced, which we named "Practicalities".
In Brazil, this cluster is related to the tour experience and unpacks the organisational details to arrange the tour, and the landmarks visited during the trip, which are placed both inside and outside the favelas. A typical text segment attributed to the second cluster and looking at the practicalities of organizing the tour is: "booking the tour was a piece of cake and basically sending a message online; the same night I got a message from Luiz with availability and confirmed through email. I was picked up from my hotel by Patrick our amazing tour guide". Travellers reporting these types of experiences come from Oceania and North America and, again, are quite experienced, aged 35+ years and mainly women. Interestingly, these contents are not linked to any specific toponymic. In India, this cluster collects terms related to the organisational aspect of travelling with a tour, such as the booking, the means of transportation, and the collection point. These practical aspects are usually associated with positive comments: "Promptly met by Suraj at church gate station; we travelled on buses, trains and taxis, taking in sights such as the Victoria terminal, the Dhobi Ghat flower and fish market and the tiffin men". The slums represented in this cluster are in Bangalore and Varanasi, with mature reviewers over 35 years and coming from North America.
South Africa and India share another theme of the slum touring experience commonly discussed in the literature, the visit to social and non-profit initiatives in the community, which we named the "Community Projects" cluster.
In South Africa, discourses in this cluster are related to a community-based type of tourism. This cluster, which is strictly linked to the eye-opening one, collects contents related to a specific non-profit initiative in Cape Town, the Uthando Project, which aims at providing a unique view into communities through responsible tourism. An example of text in this cluster is "Siviwe gave us a full look into life in the township; he shared his vision for a bright future for the people of Langa and we learned about the education traditions and culture of the supportive community". This type of content associated with slum tourism is significantly more present in reviews related to Langa, Khayelitsha, and Masiphumele (Cape Town) and to slums located near the smallest towns, such as New Brighton in Port Elisabeth. It is also relevant to notice that reviewers associated to this cluster are from another continent (Asia, Europe, North America); foreigners are still following touristic practices traditionally related to slum tourism. In India, this cluster describes discourses around non-profit projects in communities in the fields of health and education; tourists report stereotypical traits of the community, such as the happiness of children, and they describe ways to support these projects. An example of text in this cluster is "lots of kids love posing for photos, which was very sweet. The work that Pete does is so important in these slums as, without their schools, the kids would not get the free education they desperately need". Illustrative variables confirm that these terms are related specifically to the slum of Kathputli in the area of Dheli and authors associated with this cluster are males from Australia.

Destination-Specific Themes
The slum tourism experience in South Africa mediated by TripAdvisor reviews highlights two very different focuses.
The "Historical Sites" cluster is related to slum touring experiences in shantytowns in Johannesburg and, in particular, Soweto. Discourses in this cluster are structured around the historical and cultural value of these places. Touring Soweto means to tour the landmarks of the history of South Africa, especially the uprising against Apartheid and reflections on the nation's past. A usual tour of Soweto "included the Soweto stadium, the Hector Pieterson memorial, Nelson Mandela's former house, the freedom charter in Walter Sisulu Square and even a drive by of Winnie Mandela's current home". Discourses in this cluster are mainly associated with tourists from other continents, Oceania, and South America, but also from Africa and over 35 years old.
A completely opposite experience and very distant from slum touring practices reported by the literature, can be found in the cluster named "Food". Guguletu and areas in Cape Town, known as the Mother Town by South Africans, famous for its "chill out vibe", are becoming part of the city nightlife, with people going to the outskirts to enjoy some popular culture, by listening to music, dancing and relaxing with a beer, or eating good traditional home-made food. As one of the online reviewers put it, "the traditional South African home-cooked food was absolutely delicious and also quite refreshing after spending many nights at fancy restaurants in Camps Bay; the stories, music and dancing after is enough to go straight to your heart". From our data, it appears that this kind of tourism started being reported in TripAdvisor in 2012 and 2013 and the age of the reviewers is consistent with this kind of hedonistic experience, as they are aged 18-24 and 35-49 years. This atypical slum tourism experience is relevant also because, in contrast with the nationality of reviewers contributing to the community and historical clusters, reviewers significantly contributing to this cluster declare being from South Africa and Uganda and they are very plausibly from Cape Town itself. The hedonistic cluster suggests that South Africans are exploring new ways of visiting the slums, looking at these neighbourhoods not just as places to witness poverty but as parts of the city and its nightlife, producing and consuming local contemporary popular culture. It is a form of local tourism highlighting the emergence of a local class of tourists interested in exploring slums; it is a new result in slum-tourism research to be further examined in the future.
In the Brazil sub-corpus, the cluster named "Sleeping in the favela" captures a specific trait of its slum tourism as it gathers terms related to the experience of tourists sleeping over in the favelas. Brazil is the only country in our sample where reviews reported a slum tourism experience lasting more than one day, including an overnight stay within the community. The cluster related to the satisfaction of clients in terms of staff and services. "The hostel is clean and the room that we stayed in was spacious and almost entirely enclosed in windows so we could enjoy the view from the bed when we went to sleep and when we woke up on the terrace when having breakfast and all that with sounds of the ocean and birds". Reviewers contributing to this cluster are Asian and considerably younger compared to the ones attached to the previous cluster, being aged 18-24 years. Finally, these contents are typical of reviews associated with Vidigal and, to a lesser extent, to Babilonia, across the hillside in the Leme neighbourhood of Rio de Janeiro.

A Cross-Country Comparison
To sum up the results from a comparative perspective, the analysis shows that two very distinct slum tourism experiences exist in South Africa according to the location visited and they are not part of the traditional representation of slum tourism. The first one is related to historical tourism in Soweto (Johannesburg); the social media gaze here is significantly characterised by references to the main townships and to cultural and historical contents mainly related to the fights against Apartheid (in italics examples of the words significantly over-represented in each sub-corpus, p < 0.01). These include historical figures, such as Hector Pieterson, Nelson Mandela and Desmond Tutu; landmarks, such as Museums, Mandela house, Memorials and ideals and symbols, such as Freedom, Black, Protest and Struggle. The other type of experience can be labelled 'hedonistic' and linked to Guguletu (Cape Town), where an unexpectedly high number of lemmas are related to engaging with the local popular culture and cuisine traditions, including restaurants, food, buffet, lunch, meat, beer, braai, wine and vibe and shebeens (a once illegal bar that sells alcohol).
On the other hand, the results show that the experience of slum tourism in Brazil, especially in Rio de Janeiro, is already embedded into the tourism practices of the city: touring the favelas is part of visiting Rio. Lexicon relative to Brazil is significantly characterised by references not only to the favelas but also to other toponyms of Rio (Ipanema, Corcovado, etc.). This is in line with the geographical configuration of the city, in which favelas and other neighbours are intertwined; slums are not relegated to the outskirts or the periphery of the metropolitan setting. Proper nouns and nicknames of guides and locals are also very frequent (e.g., Thiago, Alberto, Olivia, etc.). Safety is an issue significantly present, as is shown by use of the terms such as safety, safe, unsafe, police, drug, to pacify, pacification, dangerous, gang, afraid. When it comes to culture, the experience of Brazil is significantly characterised by football, samba, carnival, capoeira, caipirinha, and acai. Lastly, the touristic experience in Brazilian favelas is characterised by nights spent on site, such as hostel stays, breakfast, night, bed, pousadas, dorms, and apartments.
Finally, analysis of the Indian corpus identifies all the traits of what is found in the literature of slum tourism: reviewers' reports of the eye-opening and memorable experience, mediated by a tour guide and focusing on community projects in the slum but not specific traits were identified for this country. Indian slum-tourism seems to be a 'by the book' slum experience. Terms related to industriousness are characteristics, such as industry, work, recycle, leather, profit, plastic, laundry, worker, pottery, etc. A second emerging theme is connected with mysticism and religion, as exemplified by the presence of lemmas, such as mystical, Laxmi, Shiva, Ganesh, magic, and temple. The reports bring to the foreground the poverty issue and how people cope with it (terms such as live, life condition, poverty, happiness, smile, and dweller and dirty). The anticipation of experience is linked with preconception, also derived from media (movies, photography, and Slumdog Millionaire). The reflection phase is described as a humbling and eye-opening experience, which brings with it a shock that triggers ethical considerations.

Discussion
This article aimed to contribute to our understanding of the slum tourism phenomenon in the three main slum tourism destinations (Brazil, India, and South Africa) by examining the online reviews attached to a popular tourist social media platform (TripAdvisor); therefore, investigating the social media gaze of these destinations. Particularly, leveraging the tourist gaze literature [10], as well as social representation theory [25], the study seeks to contribute to the debate on slum tourism as a peculiar form of intercultural encounter, whose negative or positive facets can be exponentially amplified by mediated communication [6].
This article gives the first extensive account of a large amount of online qualitative data, whereas the current literature usually focuses on a limited amount of tourist reports [2,90]. Results from this research both confirm the findings of previous studies and shed new light on how slums are perceived, narrated, represented, and co-constructed online by tourists, contributing to constructing tourist destinations and future representations of their inhabitants.

Shared Meanings and Variability of Slum Tourism's Representations
Cross-national analysis of specificities on the entire corpus of reports published from 2010 to 2016 confirms the differences between South African, Brazilian, and Indian experiences of slum tourism reported in the literature. Our results, based on a much more extensive dataset than previous studies, confirm that the representations of South African township tours focus on the history of apartheid and the culture of black people; references to safety and folklore characterise the favela tours; whereas industriousness and mysticism are peculiarities of Indian experiences [2]. However, when analysing each corpus separately, further peculiarities emerge.
Historical tourism in Soweto (South Africa) is only one aspect of township tourism currently occurring in the country. It seems that a new kind of slum tourism, more related to short visits to enjoy local popular culture, music, food, and dance is developing in the township constellating Cape Town. Capetonian slums are becoming locations to enjoy the city night-life and it is very interesting to notice that this kind of tourism is not performed only by the usual slum tourist (e.g., an international tourism coming, usually, from the Global North [13]). It is also enjoyed by young South Africans who are exploring new spaces of their own cities, as shown by the demographics attached to this cluster [91]. This new type of tourism emerged in the reviews of TripAdvisor from 2012-2013.
When looking at Brazil, favela discourses are characterised by violence and folk culture only if confronted with other experiences. However, when looking at the different and co-existing themes within the representations of favela tourism, a more nuanced picture emerges. From our analysis, favelas seem to be embedded in the city fabric and in the touristic offer of the city. In Brazil, this "favela normalisation" process can also be detected in the way that dwellers are very often referred to with individual nouns. In some favelas in Rio de Janeiro (e.g., Rocinha), tourists stay overnight, prolonging their stay in the favela, possibly during the visit to the city. This is supporting the integration of the favela within the tourism offer of the city.
Discourses around poverty were found in India but there were also reports on how social projects within the slums are addressing and redressing poverty and marginality. This result confirms the closeness of the Indian experience to voluntourism [92] but also suggests a possible community-based resistance to stigmatising SRs [54].

Types of E-Mediated Gaze
Regarding the types of tourist gaze, cross-country clusters confirm what is already present and discussed in the literature and can probably consider the shared processes behind the representations of slum tourism, which is beyond the specificity of each destination. The first shared process relates to the eye-opening experience of touring poverty, which fits in the literature debating the controversial nature of slum tourism and discussing tourists' gaze and motivation. It is an allocentric orientation [37], in which experience and encounter with the 'others' have a prominent role, as well as the engagement in a deep tourist gaze. This eye-opening experience is based on the 'othering' process much more than the 'sameing' one, which has been discussed in the slum-tourism and social representation literature. Our results further suggest a dynamic in the othering process: tourists oppose their experience to their expectations and shared imaginary [93]. In this sense, the representations of anticipated 'others', who were viewed as stigmatised carriers of negative stereotypes [22,94], are challenged by the experienced 'others' who are represented as a source of new insights [1].
Further research is needed to explore to what extent these changes in the representation of the 'others' deeply affect prejudices, and whether this new representation of 'others' as carriers of positive values is acceptable as far as these 'others' remain confined to their 'natural' environment. Moreover, it should be noted that in this research we examined the representation of the slum tourism in the online space. Future research is needed to examine how online and offline representations and practices interact, triangulating online data with field observation, in depth interviews or survey. These further studies are needed to understand how the self-other dynamic is activated, resisted, enacted in situ, and which role the host plays in shaping such a relationship. In other terms, further studies are needed to examine if the experience carried out by tourists and locals in the liminal spaces of inter-reality (i.e., merging online and offline, tourists, and hosts gaze) is actually an intercultural encounter or reflect multicultural expectations, which are based on essentialism and the permanence of intergroup boundaries [95,96].
For sure, this representation also protects a positive image of the tourists' in-group [54,55]. By representing slum-tourism as an eye-opening activity, the tourists justify their experience, which is rhetorically presented as a discovery of authenticity [12,13,21,84], and indirectly represents their in-group as composed of open-minded individuals. Despite these self-serving functions, it is worth noting that the described shift, from the others represented to the one experienced, contribute to defining the expectations of future tourists and may thus contribute to change in the future representations of the slums. Additionally, as 'guided tours' are the most popular way to access slums, it is possible to record how psychocentric (i.e., seeking safe experiences) and allocentric (i.e., seeking authentic experiences) gazes [37] mix in the slum tourism experience. This makes this touristic practice a space in which shallow and deep tourist gazes [43] interact and co-exist, in the same way as the "othering" and "sameing" mechanisms do.

The Role of Mediators in Shaping the Tourist Gaze and Experience
The role of the tour guide was relevant in all the corpora; the tour guide has a gatekeeping function and effectively makes the slum accessible to tourists. It is not surprising that a common theme detailing the guides and describing the guided tours in their organisational aspects has been found in this study given the nature of the platform (TripAdvisor), which aims at reviewing attractions and destinations for future tourists (the recipients of the reviews). Nevertheless, it is worth noting that this is also a positive result for reducing prejudices.
The digital re-mediation of the extended contact fostered by guides needs to be further explored in the future, but paves the way for a possible path for prejudice reduction and transformation of intergroup relationship. Indeed, guides can be listed as a social and cultural mediator of the liminal encounters between tourists and locals [46]. They are pivotal in providing a positive experience of contact that may contribute to changing the contents of prejudice for tourists who experience it personally [97]. Tour guides are the concrete Alter (other) of the interactional context [48], with whom tourists possibly negotiate alternative representations of slums and their inhabitants, and with whom locals possibly negotiate the host gaze. Once tourists report this experience online, the continuous reference to individual and positive characteristics (in terms of warmth and competence- [98]) and the use of first names could be a powerful and implicit mechanism to change the representation of slum dwellers. In fact, such a form of digitally mediated contact transforms homogenised and interchangeable others into a more variegated collection of individuals [62], including guides and experts, as well as activists, thus fostering a veritable transformation in the key aspects of intergroup encounters.

Conflicts of Interest:
The authors declare no conflict of interest.