It is always interesting to see a new approach emerge – especially in an area as long studied as kinship terminologies. Our understanding of terminological systems has benefited from the multiple and diverse perspectives represented in the analytic work of Lounsbury (Reference Lounsbury and Hunt1964a; Reference Lounsbury and Goodenough1964b), Romney (Reference Romney and Hammel1965; Romney & D'Andrade Reference Romney and D'Andrade1964), Read (Reference Read2001a), and Gould (Reference Gould2000), among others, and I hope comparable contributions will come from Jones' approach. But for that contribution to be realized some further information is needed, beyond what is in this paper. In what follows I highlight some of the major needs.

First, major innovative contributions in this area have come out of detailed formal treatments of complete systems – versus only illustrative examples. Jones does provide such an account for part of Seneca, but not for the part that is most distinctive and of most theoretical interest – the range of cousins out beyond what we call first cousins. He does address the issue, but only in a thin programmatic way – a way whose success does not seem inevitable or obvious.

Second, the Seneca detail is important because it is in that detail that Iroquois-type systems (including Seneca) differ from Dravidian-type ones. And that difference is very important because Dravidian-type terms (and thus cross/parallel determinations) map cleanly onto some social groups such as moieties, while Iroquois-type ones absolutely do not (see Lounsbury Reference Lounsbury and Hunt1964a). In this sense, “matriline” (sect. 4.1) seems much more relevant to Dravidian than to Iroquois. In a paper such as the present one, complete analyses of a number of contrasting systems is too much to ask, but one would like to have seen some illustrative detail concerning how the two types would each be dealt with, and thus the two distinguished.

Third, a similar problem is posed by Crow- and Omaha-type systems (sometimes spoken of as “skewed” systems) in which basic kinterm categories systematically cross generational lines (see Lounsbury Reference Lounsbury and Goodenough1964b). The comment about lowering the rank of DISTINGUISH GENERATIONS in note 4 does not do it, for the same reason that the treatment of Seneca cousins did not.

Finally, there is the “what does it buy you?” question. The relevance of the paper to the linguistics and universals of thought issues that open and conclude the paper depends on the success of the kin terminology analysis. What is the task whose success is at issue? That is, we already have in kinship approaches that represent semantic contrasts, approaches that model native speaker calculations, approaches that facilitate cross-system regularities and thus define types, approaches that clarify the relationship of terminologies to social features, and approaches that aid the inductive process of finding a system's regularities. What analytic or theoretical kinship questions does this new approach particularly speak to?

I look forward to seeing treatments of these issues in future publications.