But I am once again flummoxed at the implicit model of language and communication that enables conceptions such as the following:
Moreover, the expected Greek rendering of יסוד הבריאה is ἡ ἀρχὴ τῆς κτίσεως, an expression in which the word ἀρχὴ would mean "principle." It seems that this expression was changed in Mark to ἀπ᾽ ἀρχῆς κτίσεως, "from the beginning of creation" (10:6), and in Matthew further revised to ἀπ᾽ ἀρχῆς (19:4, 8). If this expression indeed underlies both CD and the sayings tradition, there must be a close genetic link between them. If this is the case, we have here a striking case in which the originality of the wording in one Gospel may be established over against that of another through comparison with the Qumranic parallel. (Kister 2009: 202; my emphasis)
Anyone familiar with redaction-critical analyses of the synoptics will recognize immediately the familiar notion of language revised in later gospels from earlier ones. So Matthew's "from the beginning" is a modification of Mark's "from the beginning of creation" rather than an autonomous instance of language that relates to the sociolinguistic phenomena of Hellenistic Greek in general. Kister takes that thinking to a new level, if I understand him rightly, by applying redaction-critical analytical tools to texts that don't exhibit a literary relationship. In other words, the comparison of Mark and Matthew at least rests on the theory that Matthew knew and copied from Mark; but what theory legitimates Kister's claim that Mark changes language from another (Hebrew) text?
The larger problem, however, stems from our theory of language and communication, if it can be said that we have one at all. As historians of antiquity, we can only work with and analyze the remains of language that was inscribed some place, whether on papyrus or parchment, on a monumental stele or statue, or even the plethora of graffiti that survives from the ancient world. It seems to me, however, that redaction criticism in general, and especially analyses such as Kister's, implicitly assume that written language was all there was. This sounds ridiculous, of course; who could imagine a silent world where people communicated solely in writing?! But this is precisely the conception of language that appears to underlie anlyses of language from text to text. Otherwise, how can Kister (or anyone else) suppose that Mark has changed "the principle of creation" to "the beginning of creation"? Why is either phrase the cause or the effect of the other?
Werner Kelber has for over two decades been arguing against just these very analytical assumptions when he refers to the "equiprimordial"-ness of variant sayings. Kelber's neologism is, perhaps, unfortunate for two reasons. First, it's just plain weird, though it seems Kelber delights in weird language (or in weirdly using language). But second—and more substantively—the concept of equiprimordiality is an exaggeration of an otherwise very insightful point. That is, Kelber argues that tradition, in oral performance, enjoys a certain autonomy vis-à-vis previous and subsequent performances of the tradition. That is, if in one performance Jesus is reported to say, "Blessed are the poor . . . Blessed are the hungry," and in another performance he is made to say, "Blessed are those who mourn . . . Blessed are the meek," the two instances are both equally ("equi-") original instances of the tradition ("-primordial"). One is not "more original" than the other. This variability in the tradition goes back as far as Jesus himself, who on different occasions may have given various renditions of the Beatitudes (cp. Matthew and Luke), and who may have sometimes correlated his Beatitudes with parallel woes (Luke) or not (Matthew).
Kelber perhaps goes too far when he insists that traditions in oral performance are equally "original" or "primordial." Tradition could certainly develop as well as devolve; it could be applied to new contexts, integrated with other traditions, or cast in new frames of reference. These would not be equi-primordial. But we should recognize the fairly obvious point that the traditions we analyze transcended and existed well beyond the texts we work with. As a result, any analysis of one text solely—or even primarily—in terms of another text runs the risk of neglecting the ways that nontextual factors contextualized both the production and the reception of our texts. In other words, in a world where people could (and presumably did) say ἡ ἀρχὴ τῆς κτίσεως as well as ἀπ᾽ ἀρχῆς κτίσεως, and many other things besides, no one was likely to perceive one phrase in terms of (or as a change from) the other.