Take any link to something with a ' in it, for instance: "Coup d'état" in the page https://en.wikipedia.org/wiki/Exile?debug=true
The API provided extract for that will be: "The 1948 Czechoslovak coup d'état (often simpl....."
The first thing that the renderer does is to html escape this. Resulting in: "1948 Czechoslovak coup d'état"
This is processed by: article.getProcessedElements(), which returns an array of strings and elements. However all these parts of the extract are handled as text content, so the entities become literal representations of the encoded entities.