Open kchall opened 10 years ago
Use information from mass pairwise FL calculations to create one version of a phoneme inventory graphic.
Suggestion from yesterday's meeting: use multidimensional scaling on distance matrices (FL, etc.) to produce a 2D graph.
For visualization tools, matplotlib is probably the easiest for command line things, but I've used pyqtgraph for plotting graphs in other QT projects. The latest stable release only supports PyQt4, not 5, but the fix to get it updated is pretty simple. The author of that package is working with a bunch of other people who work on visualization in python in general on the package called vispy, which focuses more on OpenGL plotting, so the graphics are rendered by the GPU rather than the CPU (and so are much faster and can be more complex).
Multidimensional scaling is available in scikit-learn, which is what I've used previously, and it has a ton of clustering algorithms too, which would potentially be useful for this.
create multidimensional phoneme inventories on the basis of calculated degrees of contrastiveness