Not only is featuring a very meaningful form of artistic cooperation, it can also be analyzed in a structured way. In many cases, artistic collaboration becomes in the end a piece of art, i.e. a song or an album. As countless songs are unpublished, published online on different platforms or offline in the form of single CDs or albums, the scope of this analysis has to be reduced. In order to understand more about the links in and between Algerian and Lebanese hip hop, I considered songs that featured artists across Lebanese and Algerian hop hip cultures. I only took into account songs that were released in the form of an analogue or digital album (I did not consider digital or analogue singles). Observations during field work, which I conducted for another project (Wiedemann, 2019), and I cannot elaborate on any further here, indicate that albums are considered a work of art, marking the ‘initiation’ of an artist into the world of rap. In addition to that, this approach limits the overall amount of data to a manageable amount: a database containing 165 albums by Algerian artists and 53 albums by Lebanese artists (Tables 7 and 8). These rap albums and songs include metadata on the rappers who are featured.

Information on the albums of Algerian artists was taken from a collaboratively administered and actively maintained list of Algerian rap albums on genius.com (2016a) (the last update to my database for this project was made on 7 November 2016). The people working on this list are part of an online community dedicated to rap music and can be considered experts in Algerian rap music. Their list is the most comprehensive database of Algerian rap albums I found, which makes it the best available starting point for my research. By using it, I adopt the community’s definition of who is an ‘Algerian rap artist’. Language use is not necessarily a criterion here, as the artists on this list use Algerian Arabic, Standard Arabic, French, and other languages. Neither is the artists’ nationality or self-identification as ‘Algerian’ a criterion. The main criterion is if artists resided in Algeria when they produced the album. This excludes, for example, albums produced in the diaspora by artists of Algerian nationality. It also excludes albums in which Algerian artists do not act as the host artist but are only featured. Arab diasporas would of course be a promising ground for further research (Swedenburg, 2015).

Information on Lebanese rap albums was not collected by an active online community but by myself. In my list, I tried to include all albums that were produced by rappers while they lived in Lebanon. To further research on Lebanese rap and support the community, I started a similar list on genius.com (2016b) (the last update to my database was similarly made on 7 November 2016).

The database used in this article includes the rappers’ names and aliases, and their affiliations to different rap crews, as well as their countries and cities. Some songs include more than one host artist and one featured artist. In these cases, I registered a connection between all of the involved artists. Let us take as an example the song ‘Re7la’, which was released on the mixtape ‘El 3arabi mokh’.7 It is a song uniting three well-known artists: Egyptian host artist Deeb, based in Cairo, was noticed by an international, non-Arab, non-hip-hop audience for his musical support of the Tahrir protests in 2011. One featured artist is the Palestinian, Ramallah-based rapper Boikutt, co-founder of the crew Ramallah Underground. The other featured rapper is Edd Abbas, who lives in Beirut and Abidjan and is a member of the crew Fareeq el Atrash. All of the interconnections were entered in a csv file, which was then imported into a network visualization software. I opted for displaying all connections as undirected edges between the artists. Hereby, I did not differentiate between ‘hosts’ and ‘guests’ because the artists’ roles are not always explicit in the song’s metadata. Depending on the information one wants to obtain, it is important to decide how to visualize featurings; for example, through a network in which every rapper is a node and all the rappers who appear together on one track are connected by an edge. Before visualizing the data, several decisions had to be taken.

Individual artists vs collectives: Some artists release individual work and are also active in a rap crew. Two options are possible: visualizing each individual artist (Figure 1 and Table 1) or visualizing only the crew (Figure 2 and Table 2). I opted for visualizing the crew because it is not always clear if all of the crew members took part in producing a song. In addition to that, strong ties and mutual influence between crew members can generally be assumed. Last but not least, visualizing a crew simplifies matters. Taking each individual artist as a separate node would certainly lead to a more precise picture of the network and could also take into account that line-ups of crews change and that crews can also split up.

Discretization of continuous data: Due to technical limitations, node attributes have to be reduced to discrete values despite being of continuous nature. All the artists who have lived in different countries, or traveled or commuted between different countries, or who call several countries their ‘home’ can only be placed in one country. As mentioned, I chose to pick the country were the artists lived in when they produced their work and stored these countries alongside the artists in the database.9 As my focus was on the Algerian and Lebanese rap scene and on the question of whether there is an Arab(ic) Hip Hop Nation, I opted for the following strategy: if albums were produced in more than one country and one of the artists’ countries was Lebanon or Algeria, I placed the artist there. If not, I checked if the artists’ countries included an Arab country and placed the artist there. This simplifies visualization, but information about movements of artists is lost.

The data was visualized using the software Gephi 0.9.1. I employed the Yifan Hu algorithm (for technical details, see Hu, 2006), as well as the Label Adjust algorithm, which are both implemented in Gephi. Through both algorithms, the visualization became readable as a result of the software evenly distributing nodes in space, depicting symmetry where it exists, and minimizing edge-crossing (this is at least what the Yifan Hu algorithm intends to do, see Hu, 2006: 37), as well as overlapping labels. After applying the two algorithms, the distance between the two main clusters was manually reduced and the smaller clusters were drawn nearer to the main network.

In the visualization, node and label size is based on node degree, i.e. a node’s total number of connections to other nodes. Nodes are colored according to assigned country. Those having a degree of less than one were filtered out; artists with no collaborations were not taken into account.

Even before looking at interconnections, conclusions can be drawn from analyzing the database. It contains a total of 748 artists, of whom 404 could be placed in at least one country; many could also be placed in a city. The following table (Table 3) shows the 11 cities with the highest concentration of artists within this database.

The four Algerian cities with the highest number of inhabitants – Algiers, Oran, Constantine, and Annaba – are on the list. Sidi Bel Abbès is one of Algeria’s ten biggest towns. El Harrach is a suburb of Algiers, and Boumerdès is located 50 km east of Algiers. Finding these cities in the list is not a big surprise.

Of all the Lebanese cities, only Beirut makes it to the top of the list. The Lebanese hip hop community seems to be concentrated in the capital. This might have been caused by and led to an infrastructural concentration of support personnel, production facilities, organizations, and sponsors in Beirut (Burkhalter, 2013: 63–71). Still, one should consider that many Lebanese rappers like Touffar (Baalbek), El Rass (Tripoli) and, according to the rapper Samzz, also ‘a lot of the rappers that come to Radio Beirut’ (Ajaj, 2016), are originally from other Lebanese regions.

The occurrence of three particular cities in this list is surprising: Paris, Cairo, and Ramallah are geographical hubs in the network of artistic collaboration among Algerian and Lebanese rappers. It is apparent that there are at least some artistic collaborations that cross national borders. Finding Paris on this list indicates the strong influence of Algerian and Lebanese diaspora communities.

In the network of collaborations (Figures 3 and 4), several clusters can be noticed. I analyzed if these clusters overlap with regions and countries.

There is a big cluster of mostly Algerian rappers (visualized at the top). Only a few French rappers and crews like 113, Rocé, Les Saltimbanks, and Gazateam, and Algerian artists who migrated to Europe like Intik, Africa Jungle and MBS are part of this cluster. At the bottom, we see a cluster containing many Lebanese artists but also Egyptian (Arabian Knightz, Deeb, MC Amin), Palestinian (Ramallah Underground, Al Nather), Jordanian (El Far3i, Satti), and Syrian (Bu Kolthoum, Al Darwish) artists. Several of these multinational collaborations can be found on mixtapes, which, according to DJ Lethal Skillz, ‘bridge cultures through soulful, funky, raw beats, a microphone, and two turntables’ (The Letter B, 2012). Here, the ambivalent role that international non-governmental organizations like Rapolitics, and governmental organizations like the Institut français and the United States Agency for International Development play in organizing border-crossing collaborations should be considered and examined further (Butler, 2009). Only one collaboration connects the big and mainly Algerian cluster to the big and mainly Lebanese cluster: Rayess Bek’s track ‘Al Shou3ara2’, on his album ‘Hip Hop Republic’ (2011), includes a feature by Rabah Donquishoot, one of the members of ‘Le Micro Brise le Silence’ (MBS), a pioneering Algerian rap crew.

The main take-aways of analyzing these clusters are the following. Collaboration seems to be mostly based on shared geographic location. Algerian rappers mostly collaborate with other Algerian artists and to a lesser degree with artists based in France. Lebanese artists mostly collaborate with other artists based in Lebanon, but they sometimes include artists from other Arab countries, except the Maghreb.

The degree distribution (Figure 5) produces a long tail with numerous artists (nodes) having only very few collaborations (edges) and a few artists (nodes) who have many collaborations (edges). Of all the 1,352 edges,10 the 388 artists with between 1 and 9 edges account for 211 edges. The 57 artists with more than 10 edges account for 1,141 edges. Without additional data, one can only speculate as to the reasons of this effect: Some artists might be more ‘social’ and collaborate more often than others. Especially, famous artists – all of the major hubs are well-known artists in their country – could also be more attractive partners for collaboration. Some artists have been active for a longer period of time and, therefore, have been able to collaborate with more artists than others.

Filtering out the disconnected nodes, the average number of connections of one node to others (degree) in the network is 3.038. Nine of the ten nodes with the highest degree are crews; and only one Lebanese crew is among this list of most connected artists (Table 4).

The overall network is quite fragmented: it contains many components but also several sub-clusters that are only held together by one node. This gives many hubs a broker functionality. Another indicator for the existence of brokers at a small scale is that in 669 songs, 1,500 collaborations (2.24 collaborations per song) can be found. Hence, many songs unite more than just two artists. This speaks for the broker functionality of host artists in general. Collaborations at a song-level form small cliques, as all of the participating artists are connected to each other. The detection of hubs coincides with observations from field work: While some artists’ names are not mentioned at all, others appear very often. These seem to be well-known and active artists like the rap crews Fareeq el Atrash or T.O.X who are able to initiate new collaborations.

As mentioned before, the two big clusters of mostly Algerian rappers and mostly Lebanese rappers are only connected by one edge. This edge is formed by the Lebanese rapper Rayess Bek and Rabah Donquishoot of the Algerian crew MBS. They made one track together in 2011 and also collaborated with artists of other nationalities. Meanwhile they have both migrated to Europe – Rayess Bek to Paris and Rabah Donquishoot to London. They continue to make music but their potential as brokers between the Lebanese and Algerian hip hop communities has not unfolded further; Rayess Bek experiments successfully with other styles of music whereas Rabah Donquishoot’s collaborative activities came to a halt afterwards. No indication for ongoing broker activity between the two big clusters can be found.

I focused on the micro-blogging service Twitter as it provides a useful application programming interface (API), which allows for scraping data on user interaction. The other sites have tighter API usage policies, which disproportionately complicate access to comparable data. Moreover, Twitter users generally show their profiles publicly (one can restrict the visibility of one’s Twitter profile to confirmed followers only, but this feature seems to be used very rarely), which reduces research-ethical concerns with regard to the use of Twitter data. Twitter users can post ‘tweets’, which are up to 280 characters long (before 2017, tweets had a character limit of 140), and can include photos, an animated GIF or a short video. They can ‘follow’ each other through subscribing to other users’ tweets, which are displayed in their newsfeeds. However, users who are followed by other users do not have to reciprocate the following, which makes the connections between them directed links.

Twitter can be accessed by web browser or mobile applications. Even if Twitter is called a social network, it is of course not an actual copy of the ‘real-life’ social network(s) its members live in. At best, it maps part of these ‘real-life’ networks. Not every person in a ‘real-life’ social network uses Twitter. Research in other regional contexts has shown that Twitter users are mostly younger, wealthier, and better educated than the rest of the ‘on-line population’, who are in turn younger, wealthier, and better educated than the ‘off-line population’ (Blank, 2016). In Lebanon, the hip hop community can be placed within a middle-class cosmopolitan context (El Zein, 2016). Therefore, one can assume that the presentation of Lebanese hip hop culture on Twitter is reasonably accurate. Comparing Twitter use in Lebanon and Algeria (Table 5), one notices that Twitter is used more actively in Lebanon (relative to the total population) but has probably reached its peak, whereas in Algeria, it has steadily augmented in the last couple of years.

Before starting to interpret Twitter data, it is necessary to emphasize a couple of points. Even Twitter users who follow many other users, or who have many followers, directly interact only with few contacts regularly (Huberman, Romero & Wu, 2008). This is because ‘attention is the scarce resource in the age of the web’ (Huberman, Romero & Wu, 2008: 2) and ‘the cost of declaring a new followee [sic] is very low compared to the cost of maintaining friends (i.e. exchanging directed messages with other users)’ (Huberman, Romero & Wu, 2008: 6). As a consequence, users with many followers do not necessarily interact with a lot of them (Cha et al., 2010: 13; Leonhardt, 2011). The number of followers is thus not a valid means to measure influence but it is still one possibility to measure popularity (Cha et al., 2010). Moreover, focusing on Twitter and leaving out other social media networks can only give us part of a bigger picture. Further research might concentrate on: (a) Facebook, which works as a community-building tool in both Algerian and Lebanese hip hop cultures, (b) YouTube and Soundcloud, the prime platforms through which music is disseminated.

As the previous analysis of the network of collaboration identified Fareeq el Atrash as an important hub in the Lebanese rap scene, Edd Abbas, the crew member who has the highest number of followers (1,206) and follows the most Twitter users too (932), was taken as the base for data gathering. In Algeria, Fada Vex (174 followings) of rap crew T.O.X, the most prominent hub in the Algerian cluster, was taken as the starting point for data gathering.

I used twecoll, a command-line tool written in Python, to retrieve Twitter data. For a detailed explanation of how to use twecoll, see Hammer (2016). The data was gathered starting 28 September 2016 at 19:25. Data retrieval was completed on 1 October 2016 at 09:00. One set of data was collected for each one of the artists. The collected data represents the accounts that the respective artist follows (in Figure 6: blue nodes) and the interconnections between these accounts (in Figure 6: red edges).

Here, I have to emphasize the fact that the Twitter networks are personal to Edd Abbas and Fada Vex, i.e. one does not find any accounts in them which are not connected to these artists. These networks are thus only part of a bigger network. They show only the second degrees of followings (the followings of Edd Abbas’s/Fada Vex’s followings); the two rappers themselves are not displayed.

The data gathered with twecoll were saved as a gml-file and imported into Gephi. Then, the resulting networks were laid out using again the Yifan Hu algorithm followed by Gephi’s Label Adjust algorithm. The nodes’ size corresponds to their degree. Nodes were manually colored and therefore had better readability: I opted for four (Fada Vex) to five (Edd Abbas) different sub-communities, based on a local–global differentiation, a differentiation in Twitter accounts of hip hop activists (graffiti artists, breakers, journalists, and so on), and accounts of other people and institutions.

The nodes that form the core hip hop network in Edd Abbas’ network (Figure 7) can be divided into three sub-groups: US-American hip hop, French hip hop, and Arab(ic) hip hop. However, I could not find any Maghrebis in this sub-group. In addition to these hip hop-related networks, I identified a cluster, which I call ‘Arab(ic) diverse’, which includes people and institutions active in culture, media, economics, and more, in the MENA region. This cluster also lacks Maghrebis. The last cluster ‘diverse’ is similar to ‘Arab(ic) diverse’ but it includes all the people and institutions not based in the MENA region.

Fada Vex’s network (Figure 8) was colored similarly. ‘Algerian hip hop’ replaced ‘Arab(ic) hip hop’ – there are no hip hop related MENA residents in the network besides Algerians – and the differentiation between nodes representing local and global ‘diverse’ elements was not made due to the small amount of corresponding nodes.

In contrast to the network of collaborations, these networks consist of only one giant component. This fact is caused by the data collection mechanism: only those Twitter accounts that are followed by Edd Abbas (or Fada Vex) are included and these accounts apparently are all connected to at least one other account, which is followed by the artist. What is more surprising is how dense the network is. With an average degree of 41.46 (Fada Vex: 12.74), nodes are much more densely interconnected with each other than in the network of collaborations. No clear sub-communities can be detected at first glance.

Both Twitter networks include many international rappers and hip hop activists. The biggest sub-groups consist of US-American rappers and hip hop activists. Among the latter one finds various types of rappers, from pioneers like Chuck D to mainstream artists like Snoop Dogg and Kanye West, to many ‘conscious rappers’ like Talib Kweli, Questlove, Kendrick Lamar, and Erykah Badu who tackle more sociopolitical problems in their lyrics.

Rappers outside the Arab(ic) network and the US American cluster are rare in Edd Abbas’ network. One could have also expected a strong French presence in his Twitter network due to the following reasons: a big part of Lebanese society, including Edd Abbas, are either francophone (40%) or partly francophone (15%) (Nadeau & Barlow, 2008: 311), and the French hip hop scene is the second biggest in the world (Eyre & Allers, 2013). Surprisingly, there are only a few French rappers in the network. They are not too strongly connected to the world of Lebanese rap since their degree values are quite small. Among these rappers, one finds only some well-known names like Disiz, Oxmo Puccino, Youssoupha, Akhenaton and Rim’K (who has Algerian roots). A possible explanation for the scarcity of French rappers is that in Lebanon ‘English is taking over from French’; another is that French is primarily associated with the Maronite educated elite (Miller & Caubet, 2010: 246). This could lead rappers who tackle socio-political topics to prefer other languages.

Fada Vex’s network also includes not many French rappers. Due to the cultural ties between France and Algeria and the influence French media had in the emergence of hip hop in Algeria, one would have anticipated a stronger connection. Nevertheless, the connection between the French hip hop community and the Algerian one is certainly stronger than that between the French and the Lebanese hip hop communities. The rappers with the highest amount of interconnections in this network are also those who were part of Edd Abbas’ network: Disiz, Oxmo Puccino, and Akhenaton.

While all the Arab(ic) rappers in Fada Vex’ network are Algerian, the group of Arab(ic) rappers and hip hop activists in Edd Abbas’ network is more diverse. The table below (Table 6) includes all the nodes in his network assigned to ‘Arab(ic) hip hop’ having an indegree12 of at least 100, i.e. at least 100 followers within this sample of Twitter accounts.

Table 6 proves how interwoven the hip hop scenes in different Arab(ic) countries are. One notices Lebanese hip hop artists (DJ Lethal Skillz, Eslam Jawaad, Zeinedin), but also Syrian (Bu Kolthoum, Omar Offendum, Chyno), Egyptian (Deeb, Arabian Knightz, MC Amin), Palestinian (DAM, Asifeh, DJ Sotusura), Jordanian (El Far3i), Iraqi (The Narcicyst), and Saudi Arabian (Big Hass) artists and hip hop activists. This multinational mix of artists shows that putting these artists inside country borders is not the best way to perceive the Arab(ic) hip hop scene. There seems to be one rap world encompassing rappers from Egypt, Jordan, Lebanon, the Palestinian territories, Iraq, and the Gulf States. Interestingly, no Maghrebi rappers are part of this network.