Mark: Thanks for the graphic evidence! Alas, I asked this question too late to benefit from the sale that just ended . . .
Pete: I'm using the HMT-W4 Biblica Hebraica module, and running the [NOUN gentilic] search only nets *Aramaic* gentilics. Hebrew gentilics simply aren't tagged as such--they are tagged as proper nouns. There is an earlier thread dealing with this problem: https://www.accordan...brew +gentilics; in that discussion, the existence of Hebrew gentilics is denied on morphological grounds and a workaround for identifying gentilics in the Hebrew text is proposed.
Although Hebrew gentilics are not recognized in the HMT-W4 module, they are tagged in the Qumran non-biblical mss. module (QUMRAN). Martin Abegg is credited with supervising the grammatical tagging in that module (and several others I haven't used); it would be interesting to know why he preferred to tag gentilics.
A few years ago I noticed a few oddities in the way Accordance handled Hebrew gentilics in the Biblica Hebraica module: There is an inconsistency in the treatment of some tribal names and the adjectives derived from those proper names [i.e., 'gentilics']. ‘Gad’ and ‘Gadite’ correspond to two separate Hebrew lemmata, which is correct. In the case of ‘Manasseh’ and ‘Manassite,’ however, there is only one lemma, and the four occurrences of ‘Manassite’ are tagged as the proper noun ‘Manasseh.’ The same is true in the case of ‘Levi’ and ‘Levite’: all the Hebrew forms are listed under one lemma, ‘Levi’, and since ‘Levi’ is classified as a proper noun, it is not possible to search by grammatical tag for occurrences of the plural form lewim (although it is possible to search for lewim as an inflected form). Why are these three proper name/adjective [i.e., gentilic] pairs treated inconsistently? I found that BDB distinguishes between proper noun and proper adjective [i.e., gentilic] in all three of these cases. HALOT, however, does not; it agrees with BDB in the partitioning of ‘Gad’ and ‘Gadite,’ but the four occurrences of ‘Manassite’ are tucked in at the end of the HALOT entry on ‘Manasseh,’ and HALOT likewise treats ‘Levi’ and ‘Levite’ in the same entry. I noticed on the Accordance website that in the 1998 release of the Westminster Hebrew Morphology, the lemmatization was normalized according to Koehler Baumgartner III = HALOT, which explains the difference in the tagging of ‘Gadite’ versus ‘Manassite’ and ‘Levite.’ In the case of 'Manassite,' it looks to me like an oversight in HALOT was uncritically accepted during the normalization process. In the version of HMT-W4 I'm using today (1.8), 'Manasseh' and 'Manassite' are still lumped together under the same lemma.