Let's take the proposition that traditional singers visualise the events in songs.

This has been advanced as some sort of research finding: a mediation if you like. So to research it you'd need a clear definition. Not easy to do. More than one definition involved.

What is a traditional singer, and which sort of song? Already we hit problems (how to define a trad/folk song). Then is it just narrative folk songs (ie basially ballads) or all trad songs? And then is it just folk songs or all narrative songs?

Then you have to decide about sampling = a point the Sandman has raise in the past. For me, and I only have basic social science (joint major) this is a question.

Then you have to decide how you are going to find out. You would need to choose methods, quantitative or qualitative (ie stuff you can count or more textual results). If qualitative you need a design that does what can be done to avoid biasing the outcome in terms of the questions you ask and other well-known problems.

To be fully transparent you have to account for all these definitional and methodological decisions, and, ideally, make any raw data available for people to look at as well.

On this basis, I do not agree that the work of Jim Carroll in collecting and mediating stuff about Pardon or the others has been transparent. In fact if you read his various anecdotes as told on Mudcat with care I think you can construct a persuasive argument that each time he tells them they are subtly changed to suit whatever point he wants to be making at the time. Example the one about the Irish Traveller selling songs to a printer: only in one version do we get told that the song in question (singular in that telling) was one that was well known and had featured (albeit perhaps not in the precise version sold to the Irish printer) on broadsheets before.