I've spent the day reading this study by Clinton Heylin on Bob Dylan's "House Carpenter."
Brian Peters referred to it above in this thread, and we also posted this link above. I want to recommend this piece for anybody who is seriously interested in this ballad. In attempting to answer the question: "Where did Bob get this?" Heylin launches into a major study of the history of this ballad. He examines the manuscript history and interaction, the interaction of oral tradition and the broadsides, and the relation of "The House Carpenter" to other ballads in the "Child Collection." Apparently Bob Dylan's version, which he recorded for his first Columbia album, but which was not on that album, is a rather unique version both in terms of what is usually found in North America and in terms of what was being sung during the "revival" at that time by folks like Baez, Clayton, Van Ronk and others.
Here is a sample of some of what Heylin is concerned about:
"That the De Marsan strain of 'House Carpenter' seems to have overwritten many a text that previously drew solace from British oral tradition is truly a damning indictment of what might be termed The Broadside Effect upon traditional processes. Even when there remains evidence of a British oral source underlying an American rendition, the De Marsan gloss has almost always been applied. This makes particularly problematic establishing the form and relative dispersal of texts prevalent in the US at the time of the De Marsan printing."