The concepts of mobility, tourism, transnationalism, and globalization are closely linked. Transnationalism refers to “multiple ties and interactions linking people or institutions across the borders of nation states” (Vertovec, 1999: 447) and is probably most closely linked to globalization as “complex connectivity […] the rapidly developing and ever-densening network of interconnections and interdependencies that characterize modern social life” (Tomlinson, 1999: 2). Bolander (2020: 678) comments on the difficulty of prying these two concepts apart, as they partly overlap. The differences between them depend on the conceptualization of both terms, however, and there is hence no definite answer. Some researchers have argued that transnationalism has a more limited purview (see also Bolander, 2020). Given that transnationalism is quite important for this article, it is discussed in more detail here.

Vertovec (1999) discusses six ways in which transnationalism has been conceptualized in research across different disciplines. Not all of them can be discussed or are indeed relevant here; I focus on the most pertinent aspects in the following. The most wide-spread notion is that of transnationalism as social morphology, i.e., a “social formation spanning borders” (Vertovec, 1999: 449), which is closely linked to migration and often applied in sociolinguistics (see Bolander, 2020). Tourism as the largest peaceful movement of people across cultural boundaries (Lett, 1989) can be conceptualized as a form of migration, or mobility at least (see Mohr, 2020). Thus, as people move across borders, new networks and communities of practice are formed, which constitute intriguing sociolinguistic environments worth studying. The notion of “community of practice” is usually fitting in these contexts offline but might be more complex in online spaces or at the nexus of both (King, 2019). This is elaborated on later, in Constructing Identities Online. Another conceptualization important for this paper is that of transnationalism as an opportunity for capital, in sociolinguistics often linked to commodified language practices in late capitalism (Bolander, 2020). Given that tourism is “the single largest international trade in the world” (Thurlow and Jaworski, 2017: 187), this is a central aspect in the study of the sociolinguistics of tourism and English in tourist contexts. This has been shown to be directly relevant for the physical tourist space of Zanzibar, where language use of English and the local language Kiswahili, as well as language learning are strongly commodified (Mohr, 2020; Mohr, 2021). Commodification and opportunities for capital are hence expected to play an equally central role for tourism discourse, i.e., “language and communication in tourism as global cultural industry” (Thurlow and Jaworski, 2011: 222), online. Finally, transnationalism can be conceived as hybrid cultural phenomena, i.e., “facets of culture and identity [that] are […] self-consciously selected, syncretized and elaborated from more than one heritage” (Vertovec, 1999: 451) and in that sense closely connected to transnationalism as the (re)construction of place or locality. Set in tourist spaces, the investigation presented in this paper focuses on how discourse and identity construction contribute to the construction of space, i.e., Zanzibar as a tourist destination, digitally.

Due to globalization, languages are generally less associated with nation states and “third wave” sociolinguistics focuses on speaker styles and globalized vernaculars as indexical of identities (Eckert, 2012). In World Englishes research, Bolander (2020: 682) argues, we need more investigations into “how varieties […] are used as resources for constructing place and space, particularly in […] territories where marking difference is/was coupled with drawing territorial distinctions”. This is closely linked to the notion of translingualism as employed by Canagarajah (2020: 560) for instance, which is an extension of the concept of the language repertoire, moving beyond labelled national languages. The language practices in these repertoires are then utilized to construct space, a social and affective construct as opposed to place, a primordial geographical construct (Canagarajah, 2020: 559). In the study of transnational and translingual practices, it is important to document small scale contexts and their entanglement with global dynamics (Al Zidjali, 2019). This is what the current study aims at by documenting a relatively small space, i.e., Zanzibar, and its entanglement with the global dynamics of tourism.

A move away from nationalism and towards transnationalism does not only apply on the conceptual but also on the methodological level in World Englishes. An enhanced focus on linguistics of contact and language practices would be conducive to this (Bolander, 2020) and, in line with previous research on the physical tourist space of Zanzibar (Mohr, 2019; Mohr, 2020; Mohr, 2021), this is one of the present study’s objectives as well. In this regard, the notion of the language repertoire (Benor, 2010) comprising various linguistic resources that a speaker can draw on depending on the context has been central and is again emphasized for this study.

In order to further demonstrate some of the issues raised in the previous section and to contextualize the analysis provided later in this paper, a brief description of the physical Zanzibari tourist space is necessary. Zanzibar is an archipelago off the coast of Tanzania, of which it forms part as a semi-autonomous region. Thus, it is subject to the language policies of Tanzania, with Kiswahili as the de facto national language, spoken by ca. 47 million first and second language speakers, besides the 125 other languages spoken in the country. Kiswahili is also the lingua franca of Eastern Africa, with a considerable number of speakers in neighboring Kenya, some users in Burundi, Uganda and a few in Rwanda (Eberhard et al., 2021). Unguja island, the main island of Zanzibar, is also home to Kiunguja, the standard variety of Kiswahili, and the archipelago is hence an important linguistic center in East Africa. English is the de facto national working language of Tanzania (Eberhard et al., 2021). In World Englishes research, the country has been argued to be an “Outer Circle” country where English is used as a second language (Schmied, 2008). However, it has been argued recently that English is in fact an additional language in Tanzania (for a recent account see Mohr, 2022). This seems different in Zanzibar, where tourism accounts for a major part of the local economy and many Zanzibaris learn English in order to be eligible for employment in the tourist sector (Mohr, 2020; Mohr, 2021).

Tourism has an important impact on language use on the archipelago. Thus, Zanzibaris spend a lot of financial resources on acquiring English semi-privately, i.e., in tuition classes taught by (retired) school teachers or in language classes at NGOs, given that many Zanzibaris do not think that language instruction in schools is sufficient to acquire the language fluently (see, Mohr, 2021). Besides English, many acquire additional foreign European languages like Italian or German in order to be even more attractive for the job market and to better accommodate tourists and their mother tongues (Mohr, 2019; Mohr, 2020). This closely relates to the abovementioned conceptualization of transnationalism as an opportunity for capital.

The most interesting phenomenon observed in this space and related to transnationalism as an opportunity for capital is what has been termed “Hakuna Matata Swahili” (HMS) (Nassenstein, 2019). It refers to a pidginized version of Kiswahili, allegedly developed for the tourists who usually do not speak the language (Mohr, 2020). It exhibits similarities to “mock languages”, such as Mock Spanish as described by Hill (1998) for the United States. Interestingly, the Zanzibari participants in Mohr (2019, 2020) report that it is also frequently employed for communication not only by but with tourists as it supports selling touristic goods and services. To this end, it is also printed on souvenirs, postcards etc., thus making language directly purchasable.

In view of the many languages that are spoken in Tanzania, Zanzibar is a highly multilingual space and Zanzibaris have highly diverse language repertoires. ¹ They draw on the various language practices comprised in these resources depending on different contexts and interlocutors (Mohr, 2019; Mohr, 2020). This is an illustrative example of translingual practices (see Canagarajah, 2020). When adding the tourists’ various native languages and English to this already great linguistic diversity, a “superdiverse” space (Vertovec, 2007) is created that lends itself to investigations into multilingualism and language contact. In these spaces, English can hence be considered a “multilingua franca”, i.e., one of many possible language choices in a multilingual space (Jenkins, 2015). This superdiversity is similar to digital spaces, which have been claimed to be inherently superdiverse, irrespective of tourism (see Leppänen et al., 2020). An investigation into a digital tourist space hence promises to be doubly intriguing.

Contemporary, “third wave” sociolinguistics (Eckert, 2012) is concerned with the indexicality of speech styles with regard to speaker identities. Speakers draw on various means of constructing meaning, i.e., verbal, discursive and semiotic resources, thus indexing (lack of) commonality, connectedness and groupness (Leppänen et al., 2020: 25). Nowadays, identities are increasingly viewed as temporary interactional positions “that social actors briefly occupy and then abandon as they respond to the contingencies of unfolding discourse” (Bucholtz and Hall, 2005: 591). Thus, identities are socially and situationally constructed and anything but fixed. In our globalized world, they are also complex and usually multiple, depending on individuals and groups alike. Turkle (1995: 180) mentions that “people experience identity as a set of roles that can be mixed and matched”. These dynamics are closely linked to the concept of community of practice and related notions as discussed by King (2019). Online, where there might be no shared (strong) feeling of belonging among members of a group sharing similar interests, we might rather be concerned with affinity spaces (see also Gee, 2005), as more generalized social spaces (King, 2019). Publics, i.e., indefinite, unaccountable audiences addressed through circulated discourses or identity performances (Warner, 2002), are especially relevant with regard to social media. These concepts are often interrelated and co-occurring online, with affinity spaces containing sub-groups in the form of communities of practice, for instance (King, 2019). As will be shown in the analysis, the digital Zanzibari tourist space is also complex in this regard.

Situational dependency and complexity of identities specifically hold true for online digital spaces, where sociolinguistic research into identity work is a key topic (see Thurlow and Mroczek, 2011; Tannen and Trester, 2013; Leppänen et al., 2020). There, two factors that have also been shown to be central for the sociolinguistics of tourism in general and of Zanzibar in particular, are the playfulness of identity construction and the intersection of identity work and authenticity (Thurlow and Mroczek, 2011; Leppänen et al., 2015). Playfulness refers to language use being playful in tourist contexts in order to appeal to tourists on vacation, such as HMS in Zanzibar mentioned in the previous section. Authenticity, often aimed at by stylizing the location using a local language (see Salazar, 2006 on Tanzania), is another central aspect of many tourist spaces. With regard to digital spaces, playful language use has been attested frequently online and on social media, where users draw on different semiotic resources, such as emojis, pictures and verbal language, when communicating (e.g., Thurlow and Mroczek, 2011; Thurlow and Jaroski, 2020). Authenticity plays a role for the performance of identities online in that the disembodied nature of online communication apparently makes it easier for users to express themselves in ways that would normally be perceived as inauthentic (Turkle, 1995), for example using a variety of English spoken in another country. Chong (2020) reports an example from Singapore, where young Singaporeans employ different sociolinguistic resources to construct various online personae.

As the similarities between physical tourist spaces and online spaces demonstrate, there are parallels between offline and online language use. Thus, it has been argued that we do need research into connections between offline and online discourse, as these are an important aspect of digital language use (e.g., Androutsopoulos, 2008; Leppänen et al., 2020), especially where translocally connected groups are concerned (Blommaert, 2017). This would also thwart the too narrow focus on the particularities of digital discourse as present in the beginning of the field (Mair, 2020). With regard to identity work, it has been argued that online identities are an extension of offline identities, where language choices and code-switching illustrate different facets of a person’s identities, depending on their perceived or desired social positioning (Barton and Lee, 2013: 55–68). This is particularly interesting in superdiverse spaces like the tourist space of Zanzibar (Mohr, 2020) or online spaces in general. It is this interface that the present paper aims at investigating.

The move away from methodological nationalism (Schneider, 2019) and towards a more language contact oriented, transnational methodology mentioned earlier in (World) Englishes, Mobility and Transnationalism is one of the challenges this paper aims at addressing. As mentioned above, the investigation is based on Instagram posts retrieved using geotagged data from Zanzibar, that means Stone Town on Unguja island in particular. This might at a first glance seem rather nationally oriented but the geotags were not chosen out of a concern for Zanzibar as a place where a certain language variety is spoken that needs to be studied (even if that might ultimately be an interesting issue for research in World Englishes). They were rather employed for two reasons:

a) To create a manageable window of analysis on the digital tourist space

The first issue refers to the potential size of the digital tourist space, which would be difficult to analyze in a meaningful way within a qualitative framework, at least within the scope of this paper. 30 posts were retrieved weekly from September 2019 until September 2020 ² via a dedicated API in Twitter, subsequently feeding R. This was necessary due to some difficulties with the Instagram API (see Lyons, 2019 for a more detailed discussion of this issue). Posts were retrieved from Stone Town and 5 kms around the city; an extension to other tourist locations where data for the study of the physical tourist space had been collected previously (Mohr, 2019; Mohr, 2020; Mohr, 2021) turned out to be futile due to a lack of social media activity there. Social media activity in general was limited, especially after the start of the COVID-19 pandemic and on a few occasions, there were fewer than 30 posts that could be retrieved per week.

The posts were subsequently “cleaned”, i.e., three categories of posts were excluded from the data set. This refers to 1) posts that were not cross-posted on Instagram, 2) posts from accounts that are private and 3) posts that were not related to tourism in any way. With regard to 1), retrieving data via Twitter was necessary as mentioned above. However, posts that were only posted on Twitter and not on Instagram too, i.e., cross-posted, were not relevant for the study with its focus on Instagram. Concerning 3), the criterion was not interpreted very strictly so as not to restrict the data set too much. The determining factor were the participants themselves, i.e., participant profiles and previous posts were investigated closely in order to determine whether someone might have been a tourist in Zanzibar at the time of posting. Posts by hosts were in general easier to determine. The content of the post was another criterion considered. For instance, a post by a local Zanzibari, reading only “family” and showing a family picture, was excluded from the analysis. This overall procedure left a corpus of 363 posts (out of 534) in total. However, as indicated briefly above, the number and quality of posts changed considerably after the beginning of the COVID-19 pandemic, and only the data from 2019 are hence considered here. They comprise a corpus of 205 posts with 1,842 hashtags in total. The significance and use of hashtags are outlined in the next section of this article.

Issue b) above is related to an inductive approach to data collection, which is in line with ethnography in general. The application of geotags for data retrieval did not pre-determine the choice of participants, topics discussed by the participants of the tourist space, nor did it pre-determine language practices (including scripts or other semiotic material such as emojis) chosen in that space. Searching for particular key words would have been another option for data retrieval but since I wanted to treat the nature of topics forming part of tourism discourse and hence creating the tourist space as an empirical question, I abandoned that idea. The use of key words would also have pre-determined the language practices I could have investigated and since the choice of these practices was another of the empirical questions to be answered by the study, I did not want to impact them. In this way, I applied an approach that was as inductive as possible and I pre-determined the tourist space as little as possible. This methodology then again emphasizes the close connection of place to the construction of space as suggested in transnational and translingual approaches.

Social media posts consist of many different components, drawing on diverse semiotic material (Zappavigna, 2015). In order to understand them in their entirety, all of these components have to be analyzed. Since this is the first study on digital tourism discourse associated with Zanzibar, choosing a certain focus was necessary for practical reasons. Hence, hashtags were chosen as an analytical framework based on their important functions on social media, linking different parts of the posts. Other material (text in the posts, pictures, likes) are briefly commented on where necessary. They will be the object of future analyses.

Hashtags have various functions that have been discussed extensively in the literature (e.g., Page, 2012; Zappavigna, 2015; Zappavigna and Martin, 2018). There are four functions that are relevant for this study. Others have put forth groupings into three main functions (e.g., Sykes, 2019) but a slightly more fine grained approach seemed more fruitful here. The first one is supporting visibility and findability (Page, 2012). Thus, posts can be found by other social media users interested in a specific topic and Instagram is indeed searchable for specific hashtags, which can also be followed. A user will employ a specific hashtag to be found by others interested in that topic, and possibly to generate likes. These have important phatic functions on social media in a similar way to hashtags. The visibility function is closely related to several other functions of hashtags that have been discussed in the literature. One of them is folksonomic topic marking emerging through community use (Zappavigna and Martin, 2018). This refers to users tagging their posts for what they perceive to be important issues in their posts. By doing so, they simultaneously enact the ambient community (Page, 2012), i.e., they create a subcommunity of, in this case Instagram, users interested in the same topic. Finally, they might express their attitude towards that topic or the post in general (in case it contains material they did not create themselves) and thus hashtags fulfill a meta-discursive function (Page 2012; Zappavigna, 2015). These four functions were central for the analysis provided in this article and for the coding scheme outlined in the next section.

The hashtags were coded according to several parameters using the NVivo software tool (QSR International Pty Ltd, 2020). The first parameter was language practices, i.e., English, French etc. The names for these practices were based on names that the participants in the investigation of the physical tourist space of Zanzibar used themselves (Mohr, 2019). They do not entail that I myself view them or other languages as bounded entities. For a discussion of the difficulties related to differences in epistemic interests and the choice of the “proper” ontology in sociolinguistics, see for example Røyneland (2020). Table 1 shows an overview of the language practices used in the hashtags. As can be seen there, the majority of the 1,842 hashtags (N = 1,727 = 93.8%) are in English. That means that the various and oftentimes large language repertoires of the Instagrammers are not utilized fully in hashtags in the digital tourist space of Zanzibar.

Secondly, the data was coded semantically. This resulted in 133 codes + 1 “other” category subsumed under 22 + 1 larger (“super”) categories. The most frequent of these are shown in Table 2.

An example of the abovementioned codes and super-categories is “animals” with 69 coding references, which comprised the sub-categories “cats,” “protection,” “sea creatures” and “wildlife.” Double coding was generally possible and necessary due to the complex nature of hashtags. An example of this is shown in (1), a hashtag which was coded for “feelings > positive”, “geography and places > earth”, “time” and “travel > vacation”.

1) #beautifuldestinations_earthearthfocuswonderful_placesourplanetdaily

The third coding type referred to the hashtags’ socio-pragmatic functions mentioned above, i.e., enacting the ambient community, expression of attitude, topic marking, supporting visibility/findability. Here, double coding was possible but rarer.

Like every method, this one had a few weaknesses. One issue that needs to be mentioned in this regard is that of naming language practices. While I do want to move away from conceptualizing languages as bounded entities, I did employ commonly used language names for coding. In this context, I employed these names as coding categories only. Another problem that I encountered relates to the difficulty–if not impossibility–of clearly delineating languages and attributing linguistic material unequivocally to a particular language practice in superdiverse contexts. There are several words that are frequent in tourism discourse, and could be attributed to various language practices, among them voyage, which could be French or English for example, and wanderlust, which could be German or English. In these cases, contextual decisions were made, coding the material according to the language practice used mainly in the accompanying post. This is shown in example (2) ³ , where #wanderlust was coded as English. Given the multilingual and translingual nature of the posts, these decisions are not completely incontrovertible though and eight cases that could not be determined remained (see Table 1).

2) Africa isn’t all desert […] #africa #tanzania […] #travel #wanderlust #backpacker

The other problem that I experienced was my limited language knowledge. I retrieved linguistic material in languages that I do not speak (fluently), and determined the meaning of these posts and hashtags by using Google Translate. This generally worked well but I had to ask colleagues for their support in some cases. This specifically applied to material in what turned out to be Russian but was labelled as various Slavic languages by Google Translate. This emphasizes that working on projects like these makes interdisciplinary collaboration highly desirable.

As shown in Table 2, there were six coding super-categories, i.e., larger themes, that were frequent in the data. Most of them are not surprising given the posts’ concern with tourism: they referred to geographic locations, nature or traveling, as well as social media (e.g., the insta- and -gram formations mentioned above). This is also illustrated in Figure 1, showing the 30 most frequent hashtags in the data, where #travelling, #zanzibar and #tanzania are the top three. Other hashtags noteworthy here are #sky and #sea, which are hashtags belonging to the category of “nature” and #instatravel and #instagram, which were coded for the category “social media.” The category “feeling,” which was also part of the six most frequent super-categories, is represented by hashtags such as #instagood and #love in the word cloud. ⁴

Identity was a frequently coded category; interestingly, it is only represented by #globetrotter in the word cloud. While this might imply that references to identity were not that frequent after all at a first glance, the lack of references among the 30 most frequent hashtags might rather suggest that the category as such is diverse. This is emphasized by Table 5, which shows an overview of its sub-categories and their frequency in the data.

As can be seen in Table 5, there are certain sub-categories of the “identity” category that are more frequent than others. These are “Black identity,” “girls and women,” “globetrotter,” “LGBTQ+,” “professional identity” and “travel agencies,” each with more than 10 coding references. The frequency of categories related to traveling is again not surprising given that the study is concerned with tourism discourse. Some remarks are in order though, specifically concerning the globetrotter category, which included the word itself as well as #worldtraveller. These hashtags were exclusively used by tourists, some of them travel bloggers. Their posts included several other hashtags referring to traveling and (semi)professional identities, such as blogger. An example is shown in (9). In this case, the professional identity is also related to other functions of hashtags which are important for social media personae, such as #follow and #like. These hashtags are hence central for the performance of professional identities on Instagram.

 9) T • h e • end of the road to Zanzibar TZ ^•

Hashtags such as #travel, #explore, #safari, #wanderlust, #adventure, #backpacker, #nomad, #traveller, #worldtraveller and #travelbug in (9) all refer to the Instagrammer’s identity as a traveler, as well as the appeal of traveling. This is emphasized by hashtags such as #intrepid, referring to desirable character traits for travelers, and #livelife, which stresses that traveling is a desirable activity that makes life worth living. Framing this with references to Zanzibar as a #beautiful location, even a #paradise, makes this identity even more desirable. The fact that no reference to nationality is made whatsoever (and nationality was in general not a frequent sub-category as shown in Table 5) seems to emphasize the Instagrammer’s identity as an international traveler–on their travels, nationality seems to become less relevant. This is striking, given the fact that nationalities are very much relevant when it comes to crossing borders, as certain passports allow for easier travel than others. Not mentioning nationality hence might hint more at this and other Instagrammers not needing to worry about such matters and hence creates a certain kind of in-group as well. In any case, nationality is rarely expressed in hashtags and this emphasizes the transnational nature, via a reconstruction of place or locality, of the digital tourist space. ⁵ As mentioned above, other professional identities were also sometimes mentioned. Many of these are event planners, specifically wedding planners when referring to the poster’s identity. Hashtags include #destinationpartyplanner and #nigerianweddingplannerhouston. While these hashtags certainly express identity and group membership, they also function as advertisements and thus serve visibility enhancing functions and perform the authors’ professional role on Instagram. This is the case with many of the other professional identities mentioned, including the abovementioned travel agencies, as well. Professional identities of the people posted about, for example in the accompanying pictures, are an exception. They included for example #fisherman or #musician.

The other marked identities often referred to as shown in Table 5 are Black identity, female identity and LGBTQ+ identities. These hashtags were additionally marked by using language practices other than English and especially creative hashtags, as outlined in the previous sections. These hashtags emphasize the special and sometimes minority status of these groups, at least with respect to the international travel community coming to Zanzibar. In contrast, the respective majority groups such as White travelers or men rarely indicate their (unmarked) identities by using specific hashtags (see Table 5). Further, all hashtags, except for #boy, in the category “boys and men” refer to LGBTQ+ identity as well. Examples are #gayboy and #gayguy. These seem to relate to, as mentioned in General Patterns, communities of the imagination, characterized by a strong sense of belonging or in-group membership (Wenger, 1998), and thus a marked identity. This might be one of the reasons why other language practices than English are used in the hashtags creating these communities. For people from majority groups it seems to be less relevant to express their group membership, or they indicate membership in the larger tourist community (a majority group) and affinity space less closely connected than communities of the imagination (Gee, 2005; King, 2019).

What has been shown in this section is that besides indirect means of indicating identity, i.e. using particular language practices or linguistic creativity as outlined earlier in Linguistic Patterns, there is also the direct means of indicating one’s identity by stating it in a hashtag. Neither of these means occur in isolation, they are usually combined in order for the poster to construct and perform their identity. One of the most characteristic examples of this is (4) above. The fact that all identities indicated in the examples provided here are multilayered, as illustrated for instance in (4) where Black and female identity are simultaneously invoked or in hashtags such as #gayboy where male and LGBTQ+ identities are indicated, supports previous accounts of complex identities online (e.g., Leppänen et al., 2020). It seems indeed as if roles can be mixed and matched (Turkle, 1995) in the online tourist space of Zanzibar (an affinity space), drawing on different social roles and groups (such as communities of the imagination) one feels affiliated with. The fact that many Instagrammers in the data invoked an identity as travelers also stresses that identities are temporary interactional positions (Bucholtz and Hall, 2005). The idea of traveling and the traveler inherently refer to temporariness and liminality (Mohr, 2020), as they can only be invoked in contrast to the notions of home and the people staying there. These identities are hence inherently temporary and fluid. Exceptions from this rule are the professional identities (of hosts) indicated by the hashtags, as well as the minority group affiliations mentioned above. Thus, the aforementioned Blackvoyageurs invoked their Black identity repeatedly, as did one Instagrammer from the LGBTQ+ community, using hashtags such as #gay and #gayman in all of their posts that form part of this data set. Membership in minority groups and expressing this clearly hence seems to be so central to their members that they constitute a relatively stable part of their performed identities online, even in a liminal space such as the tourist space.