As always, moving from the weighted count to the raw count just exacerbates the trends. Let’s see that broken out into topics.
Now ordinary language philosophy has a boom and bust pattern again. There are a few topics, of which it is the most prominent, where the model sees them never quite going away, but not being the center of attention once their peak passes. Oddly it also sees this pattern for wide content (which I sort of get), and causation (which I don’t). None of these see a fall in the weighted graphs, but they do in the raw graphs. I guess I think the weighted graphs are more aptly reflecting the real trends here.
It’s much easier to see the later topics on the animation.
One interesting thing that is visible here (and also in the facet graph), is that norms doesn’t really explode at the end as it does on the weighted graph. This is related to something I’ll return to in the last chapter of the book; that topic is picking up a little on changes in linguistic fashion in the literature.