Multilevel model with crossed random effects - does this model make sense?

I have a complex design and haven't found similar examples online. I'd really appreciate any views on the way I modelled it.

At a first stage, 19 editors improved the wording of 41 sentences - all editors saw all 41 sentences. At a subsequent stage, 3 judges assessed the quality of the edited sentences - all judges saw all edited sentences. I measured how editing behaviour (editing time, number of editing operations, etc.) impacts the quality of the edited sentences. So editors and unedited sentences are fully crossed at level 2, and edited sentences and judges are fully crossed at level 1 (i.e. the level at which quality measurements, the DV, are taken).

Is it OK to model this having editor, unedited sentence, edited sentence, and judge as random effects in a single mixed-effects model? Edited sentences would be nested both within unedited sentences and within editors, whilst also being crossed with judges. My intuition is that this is fine - would really appreciate your views.