Conference Agenda

"Songify Your Day" was a collaborative songwriting project led by the University of Queensland’s School of Music in partnership with the Moreton Bay Regional Council and the Community Action for a Multicultural Society program in Caboolture, a regional Queensland community. Guided by a renowned Bollywood singer and a celebrated Queensland-based composer, 25 participants from diverse cultural and linguistic backgrounds, aged 18 to 82, took part in an immersive week-long workshop. The experience culminated in a live performance featuring 13 original songs reflecting their native and English languages, cultural identities, and personal memories spanning 13 countries.

This interdisciplinary research project aimed to enhance creativity as a means of fostering human flourishing, promoting psychosocial well-being, and strengthening social engagement. Social network research (SNR) played a key role in uncovering the social dynamics intertwined with songwriting. To map these processes, we applied Relational Hyper Event Models (RHEMs), analysing group interaction and activities such as singing, movement, dancing, playing instruments, discussions, and collective decision-making. These activities were grouped into three interaction types: talking, singing, and cheering. After systematically recording who interacted with whom and when, we applied RHEMs to trace the evolving social connections that drove participants’ engagement in artistic songwriting. Our analysis accounted for hierarchical structures, pre-existing relationships, and individual psychological status measured by the Mental Health Continuum Short Form (MHC-SF). The findings provided new insights into how SNR offered valuable theoretical and analytical contributions to understanding collective music-making practices.

This study employs statistical modelling to investigate the evolution of the Australian improvised musician network over four periods (1970-1990, 1991-1997, 1998-2004, 2005-2023), with particular attention to the period following the introduction of the Creative Nation policy (1998). Using The Jazz Discography (TJD) dataset, which provides extensive data on Australian jazz musicians, the research aims to identify the processes that shape the formation of the network and, potentially, the improvisatory interactions within it. Notably, the period from 1998 to 2004 saw the highest number of recordings, which coincides with a significant restructuring of the network, possibly influenced by the policy’s impact. To model the network’s evolution, Exponential Random Graph Models (ERGM), Bayesian Exponential Random Graph Models (BERGM), and Latent Order Logistic Regression (LOLOG) models were applied. ERGM identified key network formation processes but revealed limitations, particularly in its tendency to exhibit degeneracy in triangle-based models, leading to the adoption of LOLOG. LOLOG offered a more flexible approach by accounting for network dependencies within a full probabilistic framework. Additionally, BERGM was utilized for cross-validation, ensuring that most predictors provided moderate to high acceptance with the observed network and aligned well with the prior expectations. A custom metric, “minTriadicClosure”, was developed within the LOLOG package to measure the number of nodes participating in specified minimum numbers of triadic closures among collaborating musicians, highlighting isolated trios, and network islands disconnected from the larger clusters (a common scenario present among musicians collaborating frequently). The results demonstrate significant shifts in network connectivity, especially during the 1998-2004 period, where an increase in collaborative ties might suggest that cultural and policy shifts played a crucial role in reshaping the Australian improvised music scene, though questions remain about the lasting effects of these transformations.

Music worlds and especially the dynamics of local and translocal scenes are of interest for social network scientists as well as the influence of homogamy and sometimes propinquity to understand processes of musical taste formation and transmission (Crossley et al., 2015). Among SNA researchers, while the spatiality of music worlds is considered, it is rarely the main focus of research. In computational social science, recent works are taking the spatiality of music worlds into account. Way et al. (2020) analysed an adjacency matrix between music listening practices by country and the geographic origin of the artists listened to in each country. They show that both common official language and geographic proximity between countries increasingly shape listener streams during the 2014-2019 period. Computing similarity matrices, Lee et al. (2024) found a decrease in the co-occurrence of music discovered ("Shazamed") within post-soviet republics after the outbreak of the 2022 Russo-Ukrainian War at both the country and the city levels.

In this communication, we aim at measuring the extent to which musical streaming practices vary geographically on several scales and to explore the specific case of French overseas territories. Considering that musical streaming data have become an interesting source to measure socio-cultural proximities between territories, we rely on a dataset of 2.3 B of musical streams across 16 M of users on Deezer between 2020 and 2023. First, we study the co-occurrence of listened artists across countries at the world level using similarity measures. Second, we focus on the case of French departements and compare their patterns of musical listening taking into account the role of geographic distance in the observed results. Finally, we focus on the 5 French Overseas Territories having a departemental status. Sharing a common remoteness from mainland France, they all possess dynamic music scenes and traditionnal genres of their own, whereas, at the same time, being connected to the French and transnational music industry. To better characterise the linkages between these territories, 3.1) we measure the overlap between listening practices across them and 3.2) we compare the networks of co-listened artists in each one of them. By extracting subgraphs resulting from the difference between each pairs of networks, we intend to capture local sub-structures. Finally, 3.3) we consider the songs associated to more than one artist that are listened to within these French overseas territories. We partition this network to identify a common collaboration space between the most popular artists of these territories which could partly foster the proximity between them.