Is the Penguin penalty possibly caused by anchor text on the internal site links (not external links) not being diverse enough. E.g all links point to a particular page using the same anchor text. Could this be a reason ?
I'm working on that possibility. If a site has "too many" in bound links using a certain phrase, and you also overlink using that phrase throughout your site (possibly in a sitewide navigation), then I feel like it could trip a spam filter. Some of the Penguinized sites I've looked at use the targeted phrase in sitewide navigation and/or have a dozen+ links from the homepage to internal money pages (sometimes multiple links to the same page with varied anchor text).
If you have an unnaturally high proportion of external backlinks with the same anchor text, then Penguin may devalue ALL links with that anchor text, both external and internal. If so, the loss of "juice' from the internal links would cause a bigger rankings drop than otherwise.
backlinks anchor text, keywords density, and I think that includes internal links
The difficulty is in finding the right balance, and how to steadily dumb down those effects, whilst maintaining, then building traffic back.
Really folks should not be so clinical about these factors working in isolation. A site full of backlinks covering all possibilities in many cases is not good UI. In some cases it can be. You've got to think Panda to , ie a good user experience.
But as usual I'm seeing brands getting a free ride through this, so it seems trust/brand signals cause such a weight of benefit. Nothing new here - just reinforcing what's been said before by others.