Journal reviewers (and even editors-in-chief) sometimes say things that are jaw-dropping. Here are some examples (I paraphrase here, of course):
1. The p-value (for some paired comparison) is not low enough to justify publication of the result.
2. A previous experiment showed the opposite result from yours, hence your experimental result is not credible. Variant: do a second experiment, so that our confidence in your result increases.
3. Since the authors do not use a standard method for statistical analysis, and since I do not understand the new method, I recommend rejecting the paper.
We are lucky that we psycholinguists do research that cannot result in anyone actually dying.