As usual, these things tend to be blindingly obvious once pointed out. In Guillaume Dye's fantastic critical review of the "The Qur'an in Context" (unfortunately his review is in French), he makes a point of showing that the Noldeke chronological paradigm of surah interpretation, which is followed by Neuwirth et al, cannot work because of the interpolation problem. In other words, trying to interpret the surahs as being written in chronological sequence where later surahs 'reply' to earlier surahs is impossible when one considers that the surahs were slowly cobbled together from more archaic sub-elements, including later interpolations and modifications. Thus the idea of a chronological sequence of discrete surahs is illusory, because the Qur'an was not composed in a discrete surah-by-surah fashion, with one surah finished and then the next written. Here is Dye's review (use Google translate if you cannot read French).
http://www.academia.edu/4287472/Le_Coran_et_son_contexte._Remarques_sur_un_ouvrage_r%C3%A9centDye criticizes Neuwirth's attempt to read the Qur'an's discussion of Mary in such a chronological sequence. Neuwirth argues that the "Medinan" account of Mary given in Surah 3:33-63 is a later response to the "Meccan" account of Mary in Surah 19:1-63, a replay that responds in a Medinan context to what had been previously articulated in the Meccan context of Surah 19. According to Neuwirth, the later Medinan composition revised the Mary message because it (1) 'reached out' to Christians, trying to build bridges with them; and (2) tackled Abrahamic scriptural authority more directly.
But as Dye points out, that assumes that Surah 19 was composed as an organic whole. And it certainly is not. Specifically, the virulently anti-trinitarian portions of Surah 19 are crudely interpolated into a very pro-Christian base text, and this is easily shown because they also impose a radical break on the rhyme scheme -- typically one of the most obvious signs of later interpolation. It appears that a section of anti-trinitarian couplets was jammed into what appeared to the later composers as an objectionably pro-Christian text about Mary.
Specifically, Surah 19:1-33 consists of declarations that follow the Syriac literary genre known as 'sogitha.' Statements that could have come directly from any Christian text are made throughout. At Surah 19:34, however, this genre is abandoned, as is the rhyme scheme, in a virulently anti-trinitarian section (33-40) that now uses a completely different rhyme scheme than 19:1-33 does. As Dye says, "The most logical conclusion is that it is an interpolation. In other words, the original version of Q 19:1-63 did not contain 19:34-40. From the point of view of its content, it is a text which is not anti-Christian -- it is hard to see how we could be closer to Christianity."
Dye is not the first scholar to point out that 19:34-40 is an interpolation, but he's the first to point out how this wrecks any attempt to interpret the Qur'an in the context of an assumed simple chronological transition between a Meccan and Medinan context. As he says, we actually have no idea when 19:34-40 was interpolated, where, or by whom. Just that it was interpolated to make an anti-trinitarian point. And this makes Neuwirth's attempt to analyze Surah 3:33-63 as 'Medinan rapprochement' with Christians completely useless.
Dye also rips on Neuwirth for claiming that her analysis is purely 'internal' to the Qur'an. This is like arguing that one's analysis of Qur'anic Arabic is purely 'internal' to the Qur'an -- it's a meaningless tautology. The Qur'an itself does not expressly state the circumstances of its composition; any specific answer to that question cannot be presumed 'internal' to the text, and any analysis based on such presumed circumstances is not 'internal' either -- it is based on accepting extrinsic Muslim tradition from centuries later. The fact that a scholar like Neuwirth misses this fundamental point underscores the deadening influence of the Noldekian/Muslim tradition.
Dye further notes that the one specific date we have in Qur'anic composition is, fascinatingly, for Surah 18:83-102, where Van Bladel has shown that the account of the Legend of Alexander that the Qur'an gives includes features distinctive to the Syriac apocalyptic text, composed in 629-630, that attempts to portray Heraclius' recent victories as fulfillment of the Alexandrian prophecy. The Qur'anic section of Surah 18 must have been composed after this date, since it copies -- and seemingly even approves -- Byzantine Heraclian propaganda. But this very late date for Surah 18 composition (it could have been many years after 630, but it cannot have been earlier -- the Syriac text composition gives a cutoff for the earliest date of composition, but not the latest) is impossible if the Noldekian paradigm of surah sequences is taken to reflect any real historical composition! And this very late composition date for Surah 18 (allegedly 'Meccan'!!) is totally inconsistent with traditional Muslim chronology about the time and circumstances when the surahs were 'revealed.'
Dye makes numerous other fascinating points, but I thought I'd share these few because I think he deserves to be more widely known, and also because the idea of a coherent Meccan/Medinan chronology of surah composition tends to afflict Qur'anic studies like a bad hangover ... the actual text composition being a far more complex, protracted, and multi-faceted process involving multiple 'authors.' Open question: How much of the Qur'an actually reflects Mohammed's own thoughts at all? Are all these shifting viewpoints (the original Surah 19:1-63 sans 33-40, or the 33-40 interpolation, or the 3:1-63 composition) themselves reflective of Mohammed constantly changing his mind over the years (which I think is an untenable hypothesis)? Or does just one of them reflect his 'real' view, the others being impositions by different factions who composed or modified what became Qur'anic texts? Or are all three conflicting approaches to Mary being imputed to Mo by different factions, editors, composers, and revisers over the decades -- Mo's real view on the subject being unknown to us or different altogether? These fascinating questions remain to be answered.