Applying Optimality Theory to kinship to yield the kinds of variation between and within languages found is certainly on the face of it a good idea. The OT approach can also model change (Anttila & Young-Mee Reference Anttila and Young-Mee1998), predicting which rerankings of constraints are possible diachronically, and relatedly, where two variant sets of kin equations cooccur, as in the “overlay” analysis of skewing (equations between adjacent generations; Kronenfeld Reference Kronenfeld2009; McConvell, in press).
One of the key desiderata of OT is a universal and robust set of constraints. In this article, some of the constraints incorporate questionable assumptions. For example, constraints such as Distinguish Matrikin are said to lie behind the Seneca system of classificatory kinship, with its distinction between cross and parallel kin. Distinguish Matrikin, although not rigorously defined, appears to refer to relationships of descent, specifically matrilineality.
Lounsbury (1964/1969 pp.193, 210) analysing the Seneca case, refers to the “classic but erroneous anthropological view concerning the ‘Iroquois’ type of kinship systems … that this kind of system classifies kin by membership in unilineal descent groups.” Recognition of this error, and the strong distinction then drawn by Lounsbury between “Iroquois” and “Dravidian” systems, was a breakthrough in kinship studies, further elaborated by Scheffler (Reference Scheffler, Hiatt and Jayawardena1971). It is disappointing to see a constraint apparently so closely related to this same error in this article.
Lounsbury (Reference Lounsbury and Tyler1964, p. 210) pointed out, “the classic [erroneous] theory predicts correctly only to the immediate (closest) uncles and aunts … and first cousins. Beyond this its predictions are half right and half wrong.” In this article we are initially presented with data from the immediate kin only, so the problem is not manifest. Once we go beyond that, though, we encounter contradictions between unilineal descent lines and the Iroquois/Seneca system. In an “Iroquois” system, for example, the child of female Ego's female cross cousin is classified as Ego's son or daughter, even though that child is not in the same matrimoiety or matriclan as Ego; presumably this violates Distinguish Matrikin. Child of female Ego's male cross cousin, on the other hand, is Ego's niece/nephew (brother's child).This pattern is in contradistinction to the “Dravidian” pattern of naming cross cousins' children and other relations which does match well with lineal institutions.
We see some more distant kin discussion later in the target article. The application of Lounsbury style “reduction rules” to second cousins, Older Mother's Mother's Brother's Son's Son reducing to Mother's Brother's Son, includes a reduction of MB to F, a transgression of Distinguish Matrikin and apparently ad hoc. This is part of the derivational apparatus imported to “fix up” OT constraints to generate correct terms for extended kin. The list of constraints used, however – prominently Distinguish Matrikin – may be at the root of the problem. The string “Mother's Mother's Brother's Son's Son” (MMBSS) contains the same substring “child of female Ego's male cross cousin” (MBSS) as one of the examples discussed above. Woman's MBSS, we saw, is classified as “brother's son,” in an Iroquois system, hence MMBSS is reduced to MBS without resort to replacing MB by F in the derivation.
What this points to is that the relevant constraint in an OT system must directly cope with such reductions, and not miss the mark as Distinguish Matrikin does, or have to be propped up by importation of Lounsbury-style rules. More globally, the Iroquois and Dravidian systems include many minor variants in how cross- and parallel kin are defined. There is a need for a constraint, or family of constraints, which allows for such variation to be captured by constraint rankings. The distinction between cross cousins and parallel cousins (the latter equivalent to siblings) exists in many systems, unilineal and not. In some form it is certainly a candidate for universal constraint status – the “classificatory” principle in kinship as Morgan termed it. A related constraint could be Distinguish Cross
/Parallel However variations on this compute the same-sex/opposite-sex metric in different ways. This is not the place to elaborate, but I am simply illustrating that different ways of implementing OT constraints in kinship are open.
Constraints should be made up of very basic transparent parameters and not draw in notions from outside that like descent. If this methodological principle is followed, it is then possible to compare occurrence of terminological patterns with other patterns such as lineal descent. If the two are combined in the basic mechanisms such comparisons are compromised.
Debate about the appropriateness of proposed constraints is going to be needed to lay a firmer foundation for the further development of this initiative. I would advocate disposing of any constraint like Distinguish Matrikin. Its meaning is not clear, but even if it could be clarified, the patterns it is trying to explain have nothing really to do with “matrikin.”
The sections attempting to construct a place for kinship in a very broad ontogeny of cognitive patterning are less than convincing. Claims that spatial and kinship cognition are closely related universally evidently so far rest on slim foundations of some analogous structuring. In English and other European languages there is certainly to some extent a common language for these domains, but it is not clear how widespread this is in cultures, or whether cognitive behaviour in the two domains can be shown to be parallel or sharing a common basis. Once again, however, this is a research area that offers enticing prospects.
You have
Access
Applying Optimality Theory to kinship to yield the kinds of variation between and within languages found is certainly on the face of it a good idea. The OT approach can also model change (Anttila & Young-Mee Reference Anttila and Young-Mee1998), predicting which rerankings of constraints are possible diachronically, and relatedly, where two variant sets of kin equations cooccur, as in the “overlay” analysis of skewing (equations between adjacent generations; Kronenfeld Reference Kronenfeld2009; McConvell, in press).
One of the key desiderata of OT is a universal and robust set of constraints. In this article, some of the constraints incorporate questionable assumptions. For example, constraints such as Distinguish Matrikin are said to lie behind the Seneca system of classificatory kinship, with its distinction between cross and parallel kin. Distinguish Matrikin, although not rigorously defined, appears to refer to relationships of descent, specifically matrilineality.
Lounsbury (1964/1969 pp.193, 210) analysing the Seneca case, refers to the “classic but erroneous anthropological view concerning the ‘Iroquois’ type of kinship systems … that this kind of system classifies kin by membership in unilineal descent groups.” Recognition of this error, and the strong distinction then drawn by Lounsbury between “Iroquois” and “Dravidian” systems, was a breakthrough in kinship studies, further elaborated by Scheffler (Reference Scheffler, Hiatt and Jayawardena1971). It is disappointing to see a constraint apparently so closely related to this same error in this article.
Lounsbury (Reference Lounsbury and Tyler1964, p. 210) pointed out, “the classic [erroneous] theory predicts correctly only to the immediate (closest) uncles and aunts … and first cousins. Beyond this its predictions are half right and half wrong.” In this article we are initially presented with data from the immediate kin only, so the problem is not manifest. Once we go beyond that, though, we encounter contradictions between unilineal descent lines and the Iroquois/Seneca system. In an “Iroquois” system, for example, the child of female Ego's female cross cousin is classified as Ego's son or daughter, even though that child is not in the same matrimoiety or matriclan as Ego; presumably this violates Distinguish Matrikin. Child of female Ego's male cross cousin, on the other hand, is Ego's niece/nephew (brother's child).This pattern is in contradistinction to the “Dravidian” pattern of naming cross cousins' children and other relations which does match well with lineal institutions.
We see some more distant kin discussion later in the target article. The application of Lounsbury style “reduction rules” to second cousins, Older Mother's Mother's Brother's Son's Son reducing to Mother's Brother's Son, includes a reduction of MB to F, a transgression of Distinguish Matrikin and apparently ad hoc. This is part of the derivational apparatus imported to “fix up” OT constraints to generate correct terms for extended kin. The list of constraints used, however – prominently Distinguish Matrikin – may be at the root of the problem. The string “Mother's Mother's Brother's Son's Son” (MMBSS) contains the same substring “child of female Ego's male cross cousin” (MBSS) as one of the examples discussed above. Woman's MBSS, we saw, is classified as “brother's son,” in an Iroquois system, hence MMBSS is reduced to MBS without resort to replacing MB by F in the derivation.
What this points to is that the relevant constraint in an OT system must directly cope with such reductions, and not miss the mark as Distinguish Matrikin does, or have to be propped up by importation of Lounsbury-style rules. More globally, the Iroquois and Dravidian systems include many minor variants in how cross- and parallel kin are defined. There is a need for a constraint, or family of constraints, which allows for such variation to be captured by constraint rankings. The distinction between cross cousins and parallel cousins (the latter equivalent to siblings) exists in many systems, unilineal and not. In some form it is certainly a candidate for universal constraint status – the “classificatory” principle in kinship as Morgan termed it. A related constraint could be Distinguish Cross /Parallel However variations on this compute the same-sex/opposite-sex metric in different ways. This is not the place to elaborate, but I am simply illustrating that different ways of implementing OT constraints in kinship are open.
Constraints should be made up of very basic transparent parameters and not draw in notions from outside that like descent. If this methodological principle is followed, it is then possible to compare occurrence of terminological patterns with other patterns such as lineal descent. If the two are combined in the basic mechanisms such comparisons are compromised.
Debate about the appropriateness of proposed constraints is going to be needed to lay a firmer foundation for the further development of this initiative. I would advocate disposing of any constraint like Distinguish Matrikin. Its meaning is not clear, but even if it could be clarified, the patterns it is trying to explain have nothing really to do with “matrikin.”
The sections attempting to construct a place for kinship in a very broad ontogeny of cognitive patterning are less than convincing. Claims that spatial and kinship cognition are closely related universally evidently so far rest on slim foundations of some analogous structuring. In English and other European languages there is certainly to some extent a common language for these domains, but it is not clear how widespread this is in cultures, or whether cognitive behaviour in the two domains can be shown to be parallel or sharing a common basis. Once again, however, this is a research area that offers enticing prospects.