...no single parallel has much strength (improbability) on it's own, but gains it's statistical strength as a sequential collection.
If you have a series of weak parallels that are easily accounted for by coincidence, generalization, and confirmation bias, then the end result is a weak case for intertextuality. Two wrongs don't make a right, and a dozen poor comparisons don't make a typology. The argument needs to be built on
strong cases that exemplify multiple criteria as noted in scholarship. Wishing away notable differences as evidence of those darned Flavians being oh-so clever only weakens the probability.
I should back up and say, miracle of miracles, that I somewhat agree with Alex's previous post. Demonstrating parallels between Josephus and the Gospels is not enough to demonstrate Atwill's hypothesis. The argument needs to proceed step-wise through a series of stages, and it is a
non sequitur to assume one step readily follows from the prior. Establishing that a particular gospel probably utilized Josephus does not determine that all did. Concluding that one or more gospels depend in part on Josephus does not either show that Jesus was fictional (mimesis was used even by authors of ancient non-fiction), let alone that said mythical Jesus was created by a Roman conspiracy instead of a visionary Jewish sect (the "mainstream" mythicist suggestion). Nevertheless, establishing a density of
significant correlations according to intertextual criteria is the ground floor.
But then there's also the question as to what
are the criteria for determining intertextuality. Without having some rules of thumb here, how is the argument advanced if my interlocutor simply asserts I am too "granular" and I say that he is too woolly? Granted, this is more an art form than an exact science, but the best thing to do is turn to tested and peer-reviewed guidelines. Here are a few sets of criteria, drawn from the literature review in Mary Ann Beavis, "The Resurrection of Jephthah's Daughter: Judges 11:34-40 and Mark 5:21-24, 35-43,"
Catholic Biblical Quarterly 72 (2010); 49-52:
Thomas Brodie's Criteria
1. External Plausibility - contextual factors make literary dependence probable
2. Significant Similarities in theme, pivotal clues, action/plot, completeness, order, linguistic details, complex coherence (the similarity is complex and not confused or meaningless)
3. Intelligibility of Differences
Rikki Watts' Criteria
1. Marked linguistic parallels and conceptual congruence
2. Linguistic and conceptual parallels tend toward being unique in the antecedent passage
3. Themes evoked by the allusion cohere and clarify the proceeding passage
4. Demonstration of the allusion coheres with broader themes in the recipient text as a whole
5. Similar application of the antecedent passage elsewhere
Dennis MacDonald's Criteria
1. Accessibility/availability of antecedent text
2. Analogy - tradition of imitations according to the same model
3. Density - volume of weighty similarities between the texts (a significant collection of strong parallels)
4. Sequential order of parallels
5. Shared distinctiveness of the two texts - peculiar characterization, significant word or phrase, etc.
The basic principle is that, as more criteria are satisfied, the probability of an intertextual relationship increases.
Sticking with the subject of method for now, I should note that gilius is providing a novel understanding of typology and providing criteria for which I see no evidence in the literature. Definitionally speaking, typology has been the Jewish and Christian practice of construing relationships of theological significance between key events separated in time, in which the pattern of the original or type is in some way replicated by the antitype. So, for example, Moses parting the Red Sea is typical for Joshua drying up the Jordan River, and the Sojourn and Exodus of ancient Israel is typical for the Holy Family's flight to and return from Egypt in the Gospel of Matthew. Atwill contends that there is a typological relationship between the works of Josephus and the Gospels. In his understanding, the chronologically prior events of Jesus' ministry ca. 30 CE are presented as the type and Titus' military campaign is presented as the antitype. So far so good, except of course that Atwill contends that the same circle of elites is responsible for both sets of literature. One should keep in mind that this utterly unique proposal lowers the probability of his hypothesis, because this would be the only known instance in which, in terms of actual production, the proposed type is generated after the antitype
and by the exact same community.
Furthermore, when gilius says "granular details and contexts" are not so important for typology, I'd like to know his sources for this determination. Typology is essentially a subspecies of intertextuality and thus arguments for typological connections are made stronger through multiple points of demonstrable contact with intertextual criteria. Barring a Matthean "this was done to fulfill the Scriptures" formula, you have to explore the details to make the connection. Furthermore, in the case of typology I suggest that these criteria are supplemented by attentiveness to their role in supporting coherent and allusive narrative structures. In other words, the "granular" details are necessary insofar as they suggest that the telling of the antitype as a
story is found to have new depths when compared with the telling of the type as story. Cf., among other literature, Richard B. Hays,
Reading Backwards.
What one sees in Beavis' article is an application of the criteria of intertextuality, with an eye toward narrative structures, to make her case for the Marcan passage as a typological reading of the Judges passage. She builds her case by noting common vocabulary, sequence of events, plot structure, theme, Mark's familiarity with Judges, and similar applications of the antecedent passage. Beavis also takes time to account for the differences rather than simply dismiss them as unimportant.
I've gone real meta here, but I wanted to back up and reflect on methodology a bit before going forward. This can help provide some context for everyone as we continue to examine possible parallels. I will come back in the near future to explain why I still find the specific example under discussion to be deficient.