Ao-llofams
When encountering unexplained variation in a particular language or subgroup, reconstructed PTB allofams (word families) are a vital tool for determining whether the origin of that variation can be located as far back as PTB.
The Ao languages of Nagaland (Mongsen & Chungli), for example, show intra- and inter-lingual variants of 'seven(ty)' that cannot be explained by regular sound change from a single Proto-Ao form. As it turns out, this variation can be understood as the preservation of two allofams of PTB *s-ni-s SEVEN:
< *ni | < *nis | |
---|---|---|
Mongsen | thə-ni 'seven' ni-ɹə 'seventy' |
nət-əɹ 'seventy' |
Chungli | tə́-nə́t 'seven' nə́t-ə̀ɹ 'seventy' |
In Proto-Ao, PTB final/suffixal *-s merged with *-t, and all vowels before *-t were reduced to schwa. Hence, the development of PTB *nis > PAo *nət, which is reflected in Chungli tə́-nə́t 'seven', nə́t-ə̀ɹ 'seventy', and one Mongsen variant for 'seventy', nət-əɹ.
Yet the unsuffixed allofam *ni was also preserved in Proto-Ao, yielding Mongsen thə-ni 'seven' and ni-ɹə, a variant for 'seventy'.
Such cases are a reminder of the efficacy of "taking a peek" outside a given subgroup when reconstructing (Matisoff 2003:9). No doubt more examples of such allofam preservation abound.