It seems generally agreed that questions floating about that are unlikely to be answered, in the "does X exist" category, are not especially valuable — but there are ways to improve them. In particular in such cases Skeptics.SE could add a requirement of a standard of evidence, or alternatively citing specific examples to debunk.
Flowing from a fairly lengthy but constructive chat on the topic, here is a thought that might add to the discussion, if I may quote myself:
Restricting the type of evidence is probably the better requirement, but likely to yield no responses.
So in the case of an evidentiary standard, we are no better off in the 'this question has no answers' department.
Restricting the claims to specific instances is likely to yield some answers, but there is the rather tempting follow-up: “Are these the best claims of XYZ?”
A pertinent question is which requirement ought to apply: a standard for evidence (e.g. "are there any journal articles that indicate X") or specific examples (e.g. "Is X an example of a ghost").
In reference to the original question about xenoglossia, I noted:
One distinguishing feature between the xenoglossia and the examples we have [discussed] above e.g. moon landing, dinosaurs, folding paper in half, no blue foods, is that there is no apparent or plausible theory of how xenoglossia could possibly work. Perhaps the absence of any working theory may be a useful for distinguishing claims that need a specific example and those that do not.
The caveat is that we are inviting moderators to interpret what "apparent or plausible theories" are. For example, xenoglossia is apparently the result of divine, demon, or reincarnate intervention - though those are all squarely in the supernatural category and by definition outside the scope of scientific theory.
All to say, I would agree with a restriction such as: Where there is no plausible or apparent scientific theory to explain the claim, a question must cite specific examples.
The "do/does [ghosts, demons, boogeymen, xenoglossia, gods, turtles all the way down, etc] exist" questions are generally not examined with scientific rigour. The claims are typcially waffly and the evidence unverifiable. The only way to debunk them would be to examine specific instances.
Which I say just to give air to a possible rational for choosing between requiring evidence or specific examples.
The idea behind the proposed standard is that in the absence of a theory, the only thing that can be debunked are conclusions arising out of specific instances.