"Brian, interesting point regarding the selectivity of revival... I confess I am highly partial to modal scales particularly pentatonic scales, which is probably the reason I was attracted to traditional folk in the first place."

Same here. The revival is what it is, and I'd never criticize any contemporary performer for making the same kind of choices that I have. It's more of a problem when you hear people get up and say that the folk song repertoire is full of songs about fairies etc., which creates a false impression. A lot of the magical ballads in Child are represented by a tiny number of source versions, in some cases only one (usually the famous Anna Brown).

"Do you think the earlier audio field recording collections, such as Voice of the People or US Library of Congress's archive (e.g., Lomax), are representative of the vernacular repertoire of the time, or also slanted to over-represent modal tunes and morbid themes?"

I've never checked systematically, but certainly in the case of VOTP there is a wide range of material including several CDs devoted to themes like romantic love, farm work, merrymaking, etc., and only a couple devoted to tragic ballads. I would guess that the choices there were made to showcase the best performances, and represent reasonably broadly the repertoire that those singes performed. Though you could argue that, like Sharp's collections, VOTP under-represents music hall material. If you listen to the double CD of Walter Pardon put out by Musical Traditions with the 'World Without Horses' release on Topic, you'll find more music hall and other recent songs on the MT release, which was a deliberate attempt to present a more comprehensive sample of his repertoire than the 'folksong' release had done.

"In addition to any revivalists' or collector's bias, it is also possible the singers (and indeed the "folk" in general) were more influenced by scholarly analysis and/or commercial music by the time audio recordings were widely collected (after the invention of the phonograph, and most of them post-radio) compared with the singers Sharp, Baring-Gould etc collected from."

There are examples, I believe, of singers learning ballads from earlier published material. And, in the US at least, the reason that many audio recordings of ballads made after 1930 are significantly shorter than the kind of things Sharp collected, is that those singers had learned them from 78 rpm recordings on which the ballads had been edited to fit the maximum playing time of the record.

"It seems to me (without having done statistics) that pentatonic scales are more common from Appalachian sources than European ones. Anybody know of systematic analysis of that?"

Well, there's good old Cecil himself. He was very interested in Appalachian 'gapped scales'. You can download (free) the book he co-wrote with Olive Dame Campbell in 1917, and take a look at his introduction:

English Folk Songs from the Southern Appalachians, 1917

"At the Broadside Ballad Archive project they take a stand to use only tunes that were documented (notated) before 1701, and it turns out they get most of those from fiddle or dance tune books."

If you listen to the tunes EBBA has put online, they tend to sound nothing like the tunes to which the same ballads were collected 200 years later than those Roxburghe, Pepys, etc broadsides. Which leaves the mystery of where those later tunes came from. In any case, you're certainly right to say that earlier studies downplayed the role of print, which is now much better appreciated.

"This suggests to me that many folk instrumentalists in the 1600s read musical notation but singers generally did not?"

I can't speak for the 1600s, but by 1800 (when the kind of tune books that Richard Robinson was talking about were being written out) there seems to have been a cohort of village musicians who were not members of the educated middle classes but could nonetheless read and notate sheet music (Richard, what do you think?).