In distinguishing between FACT and OPINION, Steve is demonstrating the prevailing thinking in all branches of history teaching and research. In the 1990s a number of important historians started to challenge the use of history as a nationalist tool.
Two quotes from the influential South African historian, L. van Sittert, written in 1991 demonstrate the change of thinking that the subject was undergoing. In a section of his essay called What is history? he writes:-

The tension between the elite histories of the literate professionals and the popular histories of the masses reveals “history” in all its many guises as ideology rather than fact, and always requiring critical engagement rather than mute acceptance of its claims, assertions and demands.

... and the 'critical engagement' called for evidence rather than view or opinion and the two must be clearly separated. He goes on to write:-

In the literate western tradition, up until the comparatively recent past such data was required to exist in written form to count for the purposes of historical reconstruction, but it is now widely acknowledged that oral, archaeological, linguistic and a host of other forms of data may be employed in making histories.

The debate surrounding these issues became the dominant discussion in the whole fields of researching history.
In our own little backwater, it becomes clear that the most important and influential recent book is the one by Steve Roud calling for an evidence-based approach and that evidence points towards a print-based origin for what we call traditional song - and we see this approach manifested in Steve Gardham's post above this one.
Van Sittert's call for 'critical engagement rather than mute acceptance of its claims, assertions and demands' could be interpreted for the likes of Harker to challenge in the way he did. Whether he went about this in thorough and unbiased manner is quite another matter.
I feel it is the right approach to engage written documents alongside 'oral, archaeological, linguistic and a host of other forms of data' but it seems to me that this will make it very dificult to reach a consensus.
This thread seems to be proving that.

Quotations taken from "THE MEANING AND ROLE OF HISTORY IN HUMAN DEVELOPMENT – Modern Approaches to the Teaching of History" - L.Van Sittert