all graphics by the author
In 2008, Peterson Kamwathi, a Nairobi-based Kenyan artist, was awarded a grant to visit the Dakar Biennale, in Senegal: a game changer, according to his peers. The director of the Kenyan art center that hosted his studio claimed the visit “radically changed his perspective, the way he sees things, his work, everything.”1 How did it happen, and what does such a connection entail? Looking at this event through the conceptual lens of mobility leads to explanatory models such as agency, spatial imaginary, or institutional apparatus. These are valid research domains (see Marcel 2012). However, even using comparative or multisited approaches, these methods produce analyses of single individuals, cities, or regions that do not allow us to understand more global configurations, hierarchies, and flows. A broader network is often hypothesized, seldom assessed and measured quantitatively. To what extent can social network analysis explain the geography of singular careers?
There are several reasons the network of contemporary African art is a serious subject of inquiry in the studies of art and globalization.2 Firstly, there is an overwhelmingly spatial dimension in the discourses within the field. Not only is the curatorial category of contemporary African art actually a toponym, but the spatial reference is also reiterated in exhibition titles, descriptions, etc.
Figure 1 is a remarkable example of an exhibition poster visually enacting a network across the continent, using graph aesthetics to suggest Africa is interacting with itself. This same idea of a pan-African network can be easily found in the discourses and policies of the development agencies that fund contemporary art activities in Africa. The image may not be as accurate as the discourse has made it out to be. Network analysis may help to tease out more realistic models for the distribution of African art and artists.
Secondly, there is an empirical and theoretical incentive to look deeper into what is often uncritically called the “international” or “global” scale of events or actors: convenient labels for avoiding the intricate or maybe confined geography of globalizing cities. If different parts of the African continent are indeed interacting with each other, the very existence of such a South–South network would go against what some art sociologists have called the deceptive nature of art globalization, still dominated by the “West.” Hence, accounting for this network could prove to be a decisive task to verify one of the major claims of postcolonial theory: that Europe and North America have recently gone through a process of “provincialization” (Chakrabarty 2000).
This article stems from Artl@s, a project dedicated to creating a geo-referenced database of transnational art circulations using the information contained in a written document such as an exhibition catalogue or a curriculum vitae (Joyeaux-Prunel and Marcel 2015). Using this source, I will demonstrate how the same data can be geographically disrupted in ways that create entirely different impressions about the network of African art and participating artists. To probe the contemporary African art network, I first looked at catalogues of widely recognized, field-defining events ranging from 1990 to today.3 By art event, I mean exhibitions and biennales that generate the kinds of lists (such as catalogues) that allow quantitative analysis. In such an approach, these lists serve as the evidence of connections, encounters, and mobilities.
Ethnographers may find this model difficult to swallow. For instance, anyone who has attended a biennale generally observes people ignoring each other rather than interacting, a lot like a large-scale academic conference. Besides, do the artists listed actually attend the events or is it their artwork alone that travels? In fact, this information is usually lost in catalogues. Furthermore, the little information given in these books usually results in a radical simplification of mobilities and migration histories, reducing individuals to a single (or dual, at best) origin and residence. That is why data-driven art studies can by no means replace ethnographic inquiry, but rather put forward new perspectives that can then be confronted to fieldwork evidence.
Nodes connected by links are the basic elements of a social network. In the case of an art field, what constitutes the nodes and links? The first way I conceptualize the network is to consider events as nodes that artists are connected to by attending them. This induces what could be called a “curator's network,” in the sense that the curator, who assembles the check list of participating artists, is the network generator.4
Figures 2–3 illustrate the “curator's network” approach. The layout of Figure 2 is the result of an algorithm that takes a measurement of centrality according to a node's links and places the most central nodes in the center of the graph. This provides a “distant reading” (Moretti 2013) in which we can distinguish influential/mainstream event (those in the central region of the graph), and marginal/original ones (those on the periphery of the graph). Here, the organizational details of the events are decisive: Indeed, we see the most central events are those with highly recognized curators (Simon Njami for “Africa Remix”), or powerful funders (World Bank for “Africa Now”).
Figure 3 synthesizes the first, simply replacing artist-event connexions by links between events when an artist features in several events. The layout dispatches nodes chronologically in order to visualize the circulations throughout the time frame. Interestingly, the Dak'art Biennale, which I imagined was a key node when doing fieldwork in Nairobi, appears very marginal in almost all the editions. According to this graph, if it is a network that changed Kamwathi's career, we may assume he entered by a small back door.
In order to make further geographical sense out of this network, another conceptualization of the network is possible. Figures 4–7 focus on the messier part of Figure 2, removing event-nodes all together. Instead, the nodes are now the artists that have at least two appearances in the selection of event, and links are repeated co-featurings, or instances in which both artists exhibit in the same show. This is not just a technical shift; it also completely changes the perspective on how an art network can be generated in real life. Rather than the almighty curator convening his network, this method places the emphasis on artists' practice, mobility schemes, or capabilities, suggesting they actively build preferential ties on the longer term with other artists. The operation reduces the population to 226 individuals who appear several times in the initial dataset (the “usual suspects” of Figs. 4–5).5
A lot of biennales are built following the model of the Venice Biennale, with sections sorted by nationality. In Figures 4–5, the nodes are based on these exhibition sections and locate artists with several appearances in the dataset.6 This implies something totally wrong: that an actor has a single given position in space, and that he doesn't move in time. It is nonetheless interesting as it represents an exhibited identity that certainly serves a purpose in artists' interactions and curatorial decisions. The resulting map (Fig. 4) matches quite closely the idealized network mentioned earlier, as it concentrates almost exclusively on the African continent. Some holes in the data, such as the absence of a substantial eastern or northern Africa exhibition, are worth mentioning: they could explain the relative weakness of those regions.
Although it does not resolve the problem of single-sited identities, a less ideological and probably more accurate way to locate the nodes is to use the place of residence. This produces a very different map (Fig. 5) and reveals what is actually at stake geographically in this art field. We basically see a network between four regions: Western Africa, Southern Africa, Western Europe, and Northern America. Looking at the links, representing co-featurings between two artists, and their color, indicating an average of the pair's network centrality measurements, we can also sketch some of the network's preferential circulations, that seem to map a North–South axis rather than the South–South one suggested in the introduction. Language appears to be a powerful explanation of these links, as we see a clear United Kingdom–USA–South Africa triangle echo the France–West Africa dyad.
After events and artists, a third way to generate the network is to look at cities. As the initial data set only covered ten cities, this was done by digging into the biographies of a representative selection of artists within this group,7 and geo-coding the exact longitude and latitude of each line/event archived by their curriculum vitae.
In Figure 6 it becomes clear the actors8 involved in the field of contemporary African art operate within a network of cities, rather than the simple conversation between nations that the previous illustrations may have evoked. Dakar and Johannesburg are the cities used to generate the cohort of artists, so it is unsurprising to see them stand out. However, the optimistic discourse of contemporary art “networking Africa” falls flat. Indeed, most of the contemporary African art exhibitions are in Western cities. To balance that statement, it should be said that CVs are a situated source: There is an abundance of data for central actors and vice versa. We may now ponder how relevant this network of cities is within artists' careers and lives, and whether or not artists share similar geographical stages in their progression.
In Figure 7, every dot represents an achievement in the curriculum vitae. Each node is then given an attribute, such as latitudes. This tool, inspired by the usage of prosopography by historians (Cabouret-Laurioux 2014), allows us to compare biographies and identify common features. For example, we can trace the very recent rise of southern Africa (in red, representing the latitude of South Africa alone) within artists' careers. Sorting the artists by region shows the stark difference between “local” and “diaspora” artists. Looking at this data from a statistical perspective, we can also provide solid evidence of how attending a given event changes an artist's trajectory. For example, we can state that following the first participation to the Dak'art Biennale, the proportion of sub-Saharan achievements decrease from 52% to 35% for “local artists” and from 17% to 10% for diaspora artists. This provides evidence that Dakar is not an entrance into an African network of events, but rather an exit door that artists may use as a means to emancipate themselves from African art scenes.
While some artists like Victor Ekpuk clearly used African events as a sort of launching pad, only occasionally returning afterwards, others such as Owusu-Ankomah initiate their careers in the North and later invest in Africa for very brief periods, as if to authenticate their membership in the trendy contemporary African art field. Some, like Athi-Patra Ruga, go back and forth between North and South on an almost yearly basis. In fact, there are almost as many different patterns as there are artists. This diversity in artists' trajectories and geographical patterns ultimately indicates how diverse the participants to this field really are. However, repeated encounters with artists of very different backgrounds can be analyzed as a means to expand the scope of possibilities within the network's own intricate geography. In that sense, network and biography can be seen as mutually constitutive.
We have seen that a data-driven reading of artistic events can bring forward a dispassionate understanding of events, interactions, and individual biographies. Doing that, I have tried to show both the conceptual and technical manipulation of the data that was involved, to the point we could question data-driven analysis as yet another narrative designed to fit a certain understanding of how social relations and individual motivations work. While these maps and graphs are indeed constructs, I think they should be viewed not as ultimate truths but rather as milestones in the paths to better understand contemporary African art. I see at least three such paths. First, enrich the data: The more catalogues, the bigger the numbers and the more refined the analysis. Second, work on new conceptualizations and ways to manipulate and visualize the data. Third, confront these findings to fieldwork investigation to grasp the potential meanings and power plays behind social and geographical distributions.
Interview with Danda Jaroljmek, Kuona Trust Office, Nairobi, December 2011.
Contemporary African art is not used here as an analytical category but rather as a pragmatic one. Indeed, in the past two or three decades, this terminology has been used by high-profile curators, whether for exhibitions titles or publications.
These include Dak'art 1992, 1996, 1998, 2000, 2002, 2004, 2006, 2008, 2010, 2012, 2014; “Africus” (Johannesburg) 1995; “Trade Routes” (Johannesburg) 1997; “Africa Explores” (New York) 1991; Documenta11 (Kassel) 2002; “Africa Remix” (Dusseldorf, London, Paris, Tokyo, Stockholm, Johannesburg) 2005; “Check List Luanda Pop” African Pavilion (Venice Biennale) 2007; “Africa Now” (New York) 2008; 1:54 Contemporary Art Fair (London) 2015, 2016.
It should be noted that the longest and most error-prone part in building this network was manually cleaning the database and homogenizing name spelling throughout the twenty catalogs. This was, however, a decisive step as it allowed identifying unique individuals across two separate lists.
To be more precise, this operation transforms a dataframe from a long format (a column for the events, another for the participating artists) into a matrix format (a square dataframe of 1365*1365 cells with the number of co-featurings as cell values).
Please note that the precision of the location is at national scale. I simply used the country's centroid (except for the US and Canada for visualization purposes) and a “no-overlap” algorithm to dispatch the nodes in national clusters.
My criteria were firstly to have an even geographic distribution (for both “origin” and “residence”) in Africa and in the main Western centers evidenced previously, and secondly generational distribution within the timeframe of the initial event selection. In practice, I was limited by the availability of usable, precise CVs for the batch geo-coding process I applied to the cities of each art events.
The artist selection includes Ingrid Mwangi, Owusu-Ankomah, Mounir Fatmi, Mntambo Nandipha, Dominique Zinkpe, Ablade Glover, Joel Mpah Dooh, Safaa Erruas, Victor Ekpuk, Sokey Edorh, Moataz Nasr, Myriam Mihindou, Youmbi Herve, Alleck Nirveda, Otobong Nkanga, Kiluanji Kia Henda, Peterson Kamwathi, Victor Mutelekesha, Ndary Lo, Tapfuma Gutsa, Soly Cisse, Ramanankirahina Amalia, Djibril Ndiaye, Athi-Patra Ruga, Georges Lilanga.