Visualizing musical genres as networks

By Mascha van Nieuwkerk

 

The power of genre

Genre is a powerful concept for analyzing and understanding historical concert programming. Notions of genre organize concert programs in two ways. First, they are a means of categorizing works and directing audience expectations. Second, they determine the order of the musical works on concert programs in the eighteenth- and nineteenth century. In this period programs were often based on fixed programming formats, fixed sequences of genres that were used as a blueprint for the order of musical works on a program.

Let me give a simple example to make this more concrete. Let’s say a program announces the performance of the OuvertureLa Muette de Porticiby Auber. The term ‘overture’ gives a lot of information concerning the musical work and its function on the program. The denoted work is probably symphonic (instrumentation), it makes a clear musical statement, the opening of a story, and therefore it is often performed at the opening of a concert (program function), it is probably not to long – about 5 to 10 minutes – and it is often originally part of a bigger music theatrical work (a ballet or opera) (origin). In the Dutch Concert Hall Felix Meritis the overture was a fixed item in the usual programming format. The works were regularly performed at the opening and close of the second half of the program.

Genre networks

Like every genre concept, the generic term ‘overture’ connects one artwork to other works with similar characteristics and to works with a similar function on the concert program. In the case of concert programs, these connections can get very complex because descriptions of works on concert programs often contain multiple genre concepts combining the characteristics of multiple genres in one work, or connecting multiple genres in one program item. From these connections a ‘network’ of genres emerges that reflects the local repertoire and the way it was categorized on the programs.

This historical approach to genre is radically different from a theoretical approach to genre. Whereas genre theory determines fixeddistinctionsbetween genre categories, a network analysis focuses on the connectionsbetween genres and the way these connections developover time.

Flexible dictionaries

For the Felix Meritis Concert Programs Database we have developed a toolbox to analyze the genre concepts used on the concert programs. In order to analyze genre descriptions, they should first be transformed into generic categories. For this purpose, we have created two flexible dictionaries of genre concepts. The first (key words) is a parsing method rather than a fixed dictionary. Genre concepts from the work description are translated into generic keywords. The keywords are sorted into three categories (genre; related genre and genre of original work). These categories are based on a syntactic analysis of the genre descriptions found on the programs. They relate to different constituents of the work description. Genreis the main genre mentioned in the work description; related genre refers to a genre concept that describes as a secondary characteristic of the work, genre original work is the genre of the work from which the current work originates, the work is often a fragment, an arrangement or an other type of musical adaptation:

The second dictionary (genre program function) is based on the function of a work within the program sequence. In this taxonomy classes of works that had the same function or would be interchangeable on the program are grouped as one category. For example, a ‘minuet’, a ‘polonaise’ and a ‘nocturne’ are very different genres but they all functioned as ‘short instrumental work’ presenting the instrumental soloist in the second half of the program. Similarly, vocal ‘romances’, ‘serenades’ and ‘melodies’ all functioned as ‘songs’ performed in the second half of the program before the concluding overture. It is important to note that this dictionary of program functions is only applicable to the specific, temporal and local programming practice of Felix Meritis

How to… create nodes and edges out of this

When we want to create a network of genres out of a series of concert programs, we have to make one basic assumption: genres that are mentioned in the same work description have a relationship. This is not a physical relationship, like in a network of roads or cables. It is rather a metaphorical relationship. For example, in a ‘symphonie fantastique’ there is a musical interaction between characteristics of a ‘symphony’ and a ‘fantasy’. In this case the nodes are the genres ‘symphony’ and ‘fantasy’, the edge is the relationship between those two genres.

In this case I wanted to visualize the relationships between the multiple genres mentioned in the work description, without involving the program function. Thus I needed an export of the database were these connections are listed. The connections can be found by combining the multiple lists with keywords from the work descriptions: genre work description 1, 2 and 3; related genre and genre original work.

How to…  create a graph out of this

For this demonstration I will work in Gephi. Gephi is an open-source graph visualization platform that works quite intuitively and does not need programming expertise.

–> Import the desired years via File -> Import spreadsheet. In my case, I would like to compare the genre practices under the four main conductors that were in charge of the orchestra of Felix Meritis in the nineteenth century. To be able to compare these periods I select the first seasons of every conductor. I create four workspaces:

VAN BREE 1832-1834
COENEN 1857-1859
VERHULST 1865-1867
RONTGEN 1886-1888

–> Organize your data. Chose ‘Merge into new workspace’ after selecting the years. In the Data Laboratoryyou should see all the current nodes and edges. You can rename each workspace via Workspace-> Rename

–> Calculate. In the Overviewunder Statisticsyou can apply all kinds of calculations to your network. Try Modularity

The modularity algorithm implemented in Gephi looks for the nodes that are more densely connected together than to the rest of the network (it is well explained in the paper published on the Gephi website by the guy who created the algorithm: Blondel, V. D., Guillaume, J., & Lefebvre, E. (2008) Fast unfolding of communities in large networks. Journal of Statistical Mechanics http://iopscience.iop.org/article/10.1088/1742-5468/2008/10/P10008/meta)

When you implement this measure it will determine different communities within the network, in our case, which genres are more densely connected between each other than to the rest of the network.

Set the resolution lower to get more communities (smaller ones).

–> Make a layout. The node position is random at first. Try Layout-> Yifan Huto get a better overview of the different routes within the network. You can show the different communities calculated by the modularity algorithm by using colors. ClickShow Node Labelsto see the descriptions of the different nodes. You can use Layout-> Noverlap; Layout-> Label Adjustand Layout-> Expansionto optimize the layout.

–> Color your labels. Appearance-> Nodes-> Label color-> Partition-> Chose attribute-> Modularity Class. And your nodes: Appearance-> Nodes-> Color-> Partition-> Chose attribute-> Modularity Class. Make sure the chosen palette is the same for the labels and the nodes to avoid a confusing image.

Some playing with the different visualization options will be necessary to create an uncluttered image. You should see something like this in the end:

Analysis

These visualizations give an intriguing insight in the way genre concepts were used in concert practice and the interconnectedness of these genres.

Looking at these graphs, we can make two general observations. First, it is notable how central opera is in the network in all of the four time frames and how it is connected to all sorts of instrumental genres in many different ways. This observation challenges the idea that instrumental music and opera became two separate musical spheres in the nineteenth century (Dahlhaus, 1989). Secondly, if we look at ‘concertos’ and ‘symphonies’, we see that these genres are separated from the rest of the network in the periods that Verhulst and Röntgen were in charge of the orchestra. This is interesting because these genres got a new status in this period. In the early period symphonies often had a connection to other genres such as ‘fantasies’ and ‘polonaises’, or fragments were played, like ‘adagios’ and ‘finales’. These practices gradually faded when Verhulst and Röntgen changed the programming format in order to give these works a better, more isolated place on the programs.

You can find the Felix Meritis Concert Program Database here: http://www.vondel.humanities.uva.nl/felix_new/index.php