I am an editor of a very active book series (Studies in Diversity Linguistics) myself, but I continue to write reviews for journals. However, I get more requests than I can handle, so I have introduced a principled policy, based on a journal’s approximation of best practice.
As more and more scientists are becoming aware, the situation of scientific publishing is currently a disaster. To a large extent, “publications” are behind a paywall (so they are not really public), and even when they are open access, authors have to pay (APCs), and private publishing giants like Springer or Wiley are extracting huge amounts of money from science budgets, largely because they monopolize the brands.
The only solution to this terrible situation is that academics take publishing back into their own hands, as we have done with Language Science Press, and as many colleagues have done who have invested efforts into platinum OA journals. This is what I call best practice publishing in linguistics.
But given the problems with funding platinum journals, it is clear that transition to best practice publishing cannot occur immediately. But one cost factor can be eliminated if academic linguists, in particular journal editors, work together: the cost of adapting text style to different style requirements. There is no reason why different journals should have different text-style rules, so ideally all linguistics journals should converge on a single discipline-wide text style.
If a journal that is not a platinum journal (and either works with a paywall or with APCs) helps to reach this goal by making a serious effort to help with unifying style rules, then I regard it as approximating best practice.
(in addition, journals should not list impact factors, because these are detrimental to science.)
Another policy decision that I took a long time ago is that I don’t look at the revised version of a paper whose acceptance or rejection I recommended in the first round. (In general, I think that revise and resubmit is not a good idea.)