The article, "Crossword Puzzles: Review and Implication in Urogynecology Education" by Mario Maciel de Lima, Jr (as well as Mario Maciel de Lima, and others) provides some reflections about how and why using crossword puzzles to train discipline-specific knowledge is effective. Furthermore, they cite an impressive list of publications that have backed this up. Some of the domains include anti-ulcer medication, normal birth, undergraduate pathology, medical education, pharmacology, organic chemistry, and many other domains.
I think there is nothing special about puzzles in these domains, except that they promote engaged active learning of what otherwise might be pretty dry material.
So, what if you are a teacher who wants to develop a crossword puzzle as a study aid? One thing you can do is get ahold of the hot potatoes jcross software. This allows you to create simple crossword puzzles. It won't do autofill and allow creating dense puzzles like the kind that appear in newspapers, but you can do your own layout and clues and it will create web pages that let you use them for training. This is probably fine because you expect your students to be able to look things up, and so progressive letter-hints are not as important as there being a single objective answer.
From the perspective of training, here are some things to consider:
- Be careful about ambiguity in the answers, and be sure the answers can be found in course material. For serious crosswords, the ambiguity can be fun, as long as it is resolvable. Experts are good at not prematurely answering with guesses, and so rarely have to backtrack. But novices are more likely to make guesses that are sort of correct while fitting a pattern. This can lead to frustration, and can get a solver stuck, because we are bad at identifying the errors and backtracking.
- If handled right, crossword clue/answer puzzles can often be solved repeatedly for additional benefit. For example, two different puzzles with the same answers in a different configuration can be beneficial. There is a long tradition within crossword developers of changing the difficulty with a new set of clues. Consider a second run through the same grid with harder clues.
- The self-adaptive nature of dense grids should not be ignored. Most educational puzzles are very sparse, and all clues are specific to the domain. You don't need a completely-packed dense puzzle, but adding additional 'gimme' clues (either outside the domain or within it) can make the puzzle self-adapting. That is, if a person knows the tougher answers, they will be able to solve it directly. If not, they may be able to get it after solving a few of the crossing words.
- Add additional engagement by doing time-based puzzles or team-based solutions.