I think this question is more one of orientation/model choice than code. I want to plot a correspondence analysis but I want to end up with a two-dimensional word association plot. Two examples are this, in Schonhardt-Bailey 2008:
"word distribution in correspondence space"
And this, in Anstead 2017:
Correspondence analysis showing "clusters of language use"
I have successfully run textplot_scale1d() and created a correspondence analysis, but this is a different plot to what I had in mind, because of the focus on position of speakers/documents etc (e.g. https://tutorials.quanteda.io/machine-learning/ca/ ). So I figure my problem here is about the model I'm fitting, and the model that I actually want possibly having a different name in quanteda.
I'd really welcome any help to:
(1) plot these word distribution correspondence analyses, not the positions.
And at a stretch, ideally what I would like to get to is:
(2) How to make a more advanced version of these word distribution plots, to incorporate a first step using a dictionary, so the ultimate word distribution correspondence analysis shows words that co-occur with the dictionary keywords.
Thanks very much for any advice you can offer.