In the Masoretic Text, differences between the consonantal text and the reading tradition are indicated by so-called qere notes. Here, 'what is written' (Aramaic: ketiv) does not match 'what is read' (Aramaic: qere). The different qere forms thus reflect the reading tradition. 
In the case of Biblical Aramaic, we find about 6 categories of consistent ketiv–qere mismatches. As these cannot have arisen during transmission of the reading tradition, they show that the reading tradition goes back to a different kind of Aramaic than the texts were written in.
This 'qere dialect' underlying the Biblical Aramaic reading tradition isn't just different from the Aramaic variety of the consonantal text, but it also differs from every well-attested kind of Aramaic we know.
There are some (fragments of) Aramaic texts that show a close match, but these are too short and/or of too uncertain transmission to convincingly show that they reflect the same dialect without any supporting evidence.
So let's consider the problem from the wave model of language change. This states that innovations spread over the map like ripples in a pond until they peter out. Different innovations spread differently, giving each dialect a unique combination of features.
The first feature we'll track is a conditioned change of yod to aleph. This is an innovation in the BA reading tradition compared to older Aramaic. We'll look at Aramaic varieties from four periods to see where the change started and how it spread.
In the Old Aramaic period, we have very little data to work with. Note that I've put Achaemenid Official Aramaic off to the side on this map and the next, as it bridges these periods and is as of yet impossible to localize geographically.
More data in the Hellenistic and Roman periods. *y is simply retained in most dialects. Some Qumran Aramaic documents attest a few cases of the shift we're looking for, which becomes regular in legal documents found in the Judean desert (Roman period).
To see the tail end of the waves, we'll also look at (some of) Neo-Aramaic. Most dialects preserve *y, even after thousands of years. We do find this change attested in some of the Jewish North-Eastern Neo-Aramaic dialects.
So: our change probably pops up as two separate developments. Once in Palestine, from the late Hasmonaean period onwards. And once in Mesopotamian dialects used by Jews, still attested in Neo-Aramaic. The geographical separation argues against connecting these two developments.
Next feature has to do with the assimilation or retention of *n before a following consonant, in writing at least. Origin of this feature is debated. As there are no qere notes in BA showing that the written nun should not be read, these clusters were present in the qere dialect.
Here are the attestations. Note that the Old Aramaic data is shaky: there are only a few examples or counterexamples per text, so the apparently geographical distribution could just be due to chance. In most later dialects, this seems to have been a spelling convention only.
So: this feature comes from somewhere (aliens?), spreads as a writing convention with Achaemenid Official Aramaic, and then is retained to various degrees. (Neo-)Mandaic probably doing its own thing here; not the feature we're looking for.
Third and last feature concerns a certain possessive suffix. As with the last one, the Biblical Aramaic consonantal text and reading tradition agree, even though it would have been easy to mark a difference. So the *-awhī form was preserved in the qere dialect.
Here we see typical forms developing in both Western and Eastern Aramaic. Neo-Aramaic is not included for this feature, as it has usually given up the use of this kind of suffix.
The introduction of the regional forms is pretty early, so the retention of the inherited form in the qere dialect will be useful for its classification.
Having surveyed the evidence, let's take a step back and consider the bigger picture. If we combine our features, where does the qere dialect fit in? Realistically, only two periods qualify: the Hellenistic–Roman period or Late Antiquity.
Starting with the latter, we find no place where our qere dialect would feel at home. *nC clusters are only retained in the east, but there we find either the wrong suffix or preservation of *y. In the west, *nC is assimilated and the possessive suffix is also wrong.
In the earlier period, *nC is retained everywhere, so that is not marked on the map. The east once again is inhospitable to the qere dialect, but in Palestine, we find a sweet spot: *nC retained, *-awhī largely retained and the start and peak of *āyV > *āʔV.
This is precisely the time and place that matches the 'bits and pieces' texts mentioned in the Introduction! As I will argue elsewhere in two weeks, this supports the idea that the qere dialect was a literary register used by the first rabbis.
Thank you for reading!
Thank you for reading!
For the full presentation including references, check out my Academia: academia.edu/37710126/The_o…
I'll be happy to answer your questions and hear your comments!
I'll be happy to answer your questions and hear your comments!
So I continue to be confused about which Gamliel these letters are attributed to. I was told it was Gamliel the Elder, having attributed them to Gamliel of Yavne in a previous presentation. Now in a forthcoming paper by Christian Stadel I read Gamliel II = of Yavne. Any help?
OK, Babylonian Talmud Sanhedrin 11a seems pretty clear that it was Gamliel of Yavne. Not Gamliel the Elder as written on my slide. I probably misunderstood a comment at some point.
