Greg Lindahl's comments are pretty accurate. The mistake people have tended to make is to assume that supernatural elements in a song 'must' somehow indicate antiquity. What they actually indicate is period; and, in song, the supernatural has been fashionable at various periods; chiefly (from the records we have available) the 18th and 19th centuries.

That doesn't necessarily invalidate the comments made here, but it does reinforce the point (frequently not grasped by amateur folklorists, who tend to rely on outdated authorities) that any analysis needs to be based on concrete evidence rather than mere 'intuitive' speculation. Wimberly and his contemporaries tended to follow Frazer's theories, but folksong scholarship has moved on a wee bit over the last century, just as it has in other fields.