Oral performance compared to manuscript versions
Professor Daron Burrows sings Laisse 1 of the Chanson de Roland.Scholarly consensus has long accepted that the Song of Roland differed in its presentation depending on whether transmission was oral or textual; namely, although a number of different versions of the song—containing varying material and episodes—would have been performed orally, the transcription to manuscript resulted in greater cohesiveness across those versions.
Early editors of the Song of Roland, informed in part by patriotic desires to produce a distinctly French epic, could thus overstate the textual cohesiveness of the Roland tradition. This point is expressed by Andrew Taylor, who notes, "[T]he Roland song was, if not invented, at the very least constructed. By supplying it with an appropriate epic title, isolating it from its original codicological context, and providing a general history of minstrel performance in which its pure origin could be located, the early editors presented a 4,002 line poem as sung French epic".[7]
AOI
Detail of manuscript showing "AOI" at the end of the second lineCertain lines of the Oxford manuscript end with the letters "AOI". The meaning of this word or annotation is unclear. Many scholars have hypothesized that the marking may have played a role in public performances of the text, such as indicating a place where a jongleur would change the tempo. Contrarily, Nathan Love believes that "AOI" marks locations where the scribe or copyist is signaling that he has deviated from the primary manuscript: ergo, the mark indicates the source is a non-performance manuscript.[8]