Poet-Composer Collaborations in Egyptian Song: A Social Network Analysis Approach to Egypt’s Musical History
Michael FrishkopfThis paper presents the results of an unprecedented application of social network analysis to a large Egyptian song database, revealing patterns, that would otherwise pass unnoticed, in the network of creative poet-composer collaborations at the heart of Arabic music. Our analysis centers on an early 21st century listing of songs played on Egyptian national radio. This database comprises 18,591 rows, each representing one song and indicating title, composer, and poet. Each song implies a collaboration between a poet and a composer. The entire database can then be interpreted as a two-mode network: each artist (poet or composer) is a node, and the number of songs produced by a particular poet-composer pair (representing the strength of their artistic relationship) is the value of the link connecting them. We seek to understand the patterns, local and global, inherent in these collaborations, via analysis of degree, centrality, and cohesions, and to interpret them by reference to cultural and historical context. We compute eigenvalue centrality of each node, recursively defined as the sum of the centralities of its neighbors, weighted by the number of collaborations. Thus node centrality depends not only on the number of collaborators or collaborations, but also on the centrality of its collaborators (a concept similar to that of Google’s Page Rank algorithm.) Using powerful social network software (Pajek), we conducted an algorithmic analysis of the full network, comprising 3160 artists and 6727 links, thereby answering questions about Egypt’s musical history that could not be addressed (and might not be asked) without the social network analysis approach. Among other conclusions, we highlight the most productive collaborations in Egyptian musical history, relating them to cultural and historic context. We demonstrate that the average composer works with twice as many poets as poets work with composers, and collaborates on twice as many songs. We highlight the 100 top (most central) poets and composers, showing that they constitute a densely interconnected, cohesive group. Moreover, the top 10 composers work with three times as many collaborators as top 10 poets, while top poets are twice as likely to work with top composers than top composers are to work with top poets. Of the cohesive core (131 nodes) obtained by eliminating nodes that have collaborated on fewer than 8 songs, 74% is connected. In sum, the average poet collaborates with fewer composers, and is less musically productive than the average composer; subnetworks comprising the most central or collaborative artists are densely connected, implying homogeneity of the core Egyptian musical style. The paper will elucidate aims and methodology, present analytical conclusions and interpretations, and propose future directions in the applications of social network analysis to Arabic song, now arising through the increasing availability of large song databases.