The narrative tradition, with the exception of the imaginative, legendary haggadah type, was mainly formulated when the eyewitnesses at a later point needed to illustrate a particular question that was posed to them. (8; my emphasis)
If I'm reading him rightly, Byrskog seems to suggest (or to say that Gerhardsson believes) that rabbinic haggadic tradition stems from eyewitness memory unless it exhibits "imaginative, legendary" characteristics. This strikes me as an astonishingly nineteenth-century thing to say, as if critical scholars can accept without worry traditional materials from antiquity that transfer relatively easily into an Enlightenment framework and can discard those materials that violate Enlightenment sensibilities.
I'm not sure this approach to ancient texts can be sustained any longer. Instead, the memorial dynamics that produced more mundane "narrative tradition" also led to "the imaginative, legendary haggadah type" of Jewish tradition. Both stem from how ancient tradents (i) understood the worlds in which they found themselves and (ii) navigated the texts that explained and programmed their experiences in those worlds. These memorial dynamics are thoroughly social, and I've criticized Byrskog elsewhere on precisely this point (see chapter 3 of Structuring Early Christian Memory). Eyewitnesses at some later point may have needed to formulate accounts of their past in order to answer this or that question, but Byrskog's analysis needs to address the social factors that determined how those eyewitnesses formulated their accounts.