I would still like to know what generally accepted definition Jim claims to have been changed by Steve Roud's book.

The collectors around 1900 knew a folk song when they heard one; or rather they knew what they thought qualified as a folk song and so was worth collecting. Some of them were mainly interested in the tunes, so didn't care if the words had been printed on a broadside, let alone whether the broadside was the first incarnation or itself based on previous oral tradition.

Jim tells us that Walter Pardon et al knew which of their personal repertoires they regarded as folk songs and which were something else. But how did they decide? What criteria did they apply, consciously or instinctively?

Steve G says, "the 54 thing is NOT a definition, it is a list of descriptors to enable scholars to identify what might be folk songs."

The book presents a somewhat larger set of descriptors but Steve R states that he more or less agrees with "1954".

If we take this part of "1954" - "music which has originated with an individual composer and has subsequently been absorbed into the unwritten living tradition of a community" - and qualify it by "where the individual composer belonged to the common people and did not compose primarily for payment", would that capture Jim's idea?

Did anyone previously include that qualification or something like it?