Matthew's accounts of Jesus' childhood set the stage for Jesus' ministry depicted in the rest of the Gospel, "defining his origin and goal" (Meier 1980:1-2). Matthew builds almost every paragraph following the genealogy and preceding the Sermon on the Mount around at least one text of Scripture. He thus invites us to read Jesus in light of Scripture and Scripture in light of Jesus-to recognize that the person and work of Jesus are central to Scripture's character. Some have suggested that Matthew made up the infancy stories to fit Scripture texts about Jesus, but the evidence suggests that he instead chose the Scripture texts to fit the stories. Matthew hardly cites the most obvious messianic texts here (see Soares Prabhu 1976); he probably depends on earlier tradition for the stories (Davies and Allison 1988:190-95; see more fully France 1981).
Because skeptical scholars often challenge the historical accuracy of these stories, we pause to consider some responses here. First, because Matthew builds each paragraph around Scripture, it should not surprise us that the passages contain midrashic (Jewish interpretive) elements, but we should not a priori assume that these passages are therefore necessarily unhistorical (see Hagner 1993:lviii; against Bourke 1960; Beare 1981:82). While midrashic interpretation can lead to fanciful elaborations of biblical accounts, it need not do so; at its core midrash simply requires reflection on Scripture. It is thus too simplistic to simply define Matthew's narratives as midrash and to define midrash as historically inaccurate embellishment (see P. Payne 1983; Cunningham and Bock 1987).
Second, were Matthew embroidering the stories of Jesus' birth, one might expect him to do so far more thoroughly, as various Jewish accounts embroidered biblical births (such as 1 Enoch 106:2-3) or later apocryphal infancy Gospels embroidered Jesus' childhood.
Third, many texts Matthew cites are hardly obvious ones if he started with texts and then created stories around them (the stories do not fit the texts as well as one would expect if he had started with the texts first). While ignoring obvious messianic texts like Isaiah 9:6, Matthew selects texts that his contemporaries did not regard as messianic (Gundry 1975:194, despite his later view). This suggests that Matthew started with the infancy traditions and found biblical texts that fit (albeit adjusting the telling of both in the process). As R. T. France observes, "The only conceivable reason for introducing these texts is that it was already known that Jesus went to Egypt, that there was a slaughter of children, and that Jesus' home was in Nazareth, and that scriptural justification was desired for these elements in his background" (France 1985:71).
Fourth, scholars concerned to trace each detail of their narrative to extant sources can easily abuse midrash. Whereas some Matthew scholars suggest that some of Matthew's infancy material is midrash on traditions later incorporated into Luke (for example, Gundry 1982:20), some scholars who specialize in Luke suggest that Luke has midrashically interpreted Matthew (Drury 1976:123-25).
Fifth, the infancy narratives' vocabulary is not much more characteristic of Matthew than the rest of the Gospel is (Soares Prabhu 1976:167-70). Indeed, substantial evidence supports the possibility that earlier narratives stand behind Matthew's infancy stories (Davies and Allison 1988:190-95).
Finally, midrashic elements do not make these accounts strictly midrash in themselves (see Wright 1966:454-56). Given the literary unity and Luke's occasional confirmation of Matthew's birth narratives apart from the Old Testament citations, Matthew probably added the quotations to the narratives rather than simply the reverse (Down 1978; also see Soares Prabhu 1976:159-60, 165).
Had Luke reported the same undoubtedly pre-Matthew traditions, many scholars would be inclined to accept their historical core on the basis of multiple attestation or at least a pre-Synoptic tradition. Yet Luke's and Matthew's divergences do help us historically in one way: because these infancy narratives are independent (so also, for example, Hagner 1993:14-15), the points where they do overlap provide independent attestation for pre-Synoptic tradition-such as the virgin birth.