As we continue to work through the geometric morphometric data, I have been exploring potential methods of synthesizing the various qualitative attributes collected over the course of that analysis. Over the previous few months, I have been building networks using the qualitative attributes associated with the scanned vessels to explore potential connections between various vessel forms, types, tempers and firing practices, and am currently working to incorporate burial data.
In the network above, red nodes correspond with vessel forms and blue nodes with vessel types; each is sized proportional to their contribution to the overall assemblage. Once built, the network can be analyzed using a variety of methods including modularity (below).
Modularity measures how well a network decomposes into modular communities. In this instance, the algorithm was randomized to produce a better decomposition resulting in a higher modularity score. The modularity algorithm looks for nodes that are more densely connected together than the rest of the network, and the colors indicate different communities determined by the algorithm. In this case, it shows which forms/types are more densely connected between each other than the rest of the network.
These networks and analyses may help to further current dialogues in the ancestral Caddo region; particularly those associated with more qualitative pursuits. Moving forward, it may be fruitful to generate these networks using variable spatial scales and temporal ranges.
The complexity of the networks will vary dramatically, depending on the number of vessels recovered. This becomes much clearer when a site with a large number of vessels (Tuck Carpenter; above) is contrasted with a site with fewer vessels (George C. Davis; below). In the example below, both the nodes and the text are scaled proportional to their contribution to the overall sample.
Due to the nature of networks and the often variable detail of those data used in the analysis (i.e., whether the analyst used type varieties, form numbers, etc.), much of the data must be revised prior to incorporation into a larger, more synthetic network (i.e., for analyses associated with Phases, spatial/temporal ranges, etc.).
The analytical power of network analysis is evident within the graphs; however, much work remains ahead of us as we contemplate how to best synthesize these data within the framework of broader research questions. For instance, how do vessel form and type vary between the currently-defined Caddo Phases, and how well do these networks correlate with our current interpretations?
More to come.