[Note: this transcription was produced by an automatic OCR engine]

43

section system tenninology. However, it will be necessary to study its replacement
by another category in Ooldea 2 and among the Mandjindja.

' Panaka is mostly combined with Paljeri. When Paljeri is used, so is Panaka, with the
exception of two groups.

° Milangka appears jointly with Tjarurru in four groups out of seven. Milangka and
Tjanrrru are, therefore, not substitutable.

' Yiparrka appears jointly with Tjarurru in five groups out of five. Yiparrka is, therefore,
not a substitute for Tjarurru.

0 The number of combinations of Burgulu with other categories reflects its dislocated
character. Except for the combinations with Purungu and Ka.ri.ma.rra, which may be
explained by the generalised character of the latter, Burgulu does not have a preferred
combination with another category. Although used by four groups, it is combined
with five different categories.

The lack of combination or joint appearance of two categories, the complementary
combinations, will permit a preliminary formulation of general rules of substitution.
Following is the list of complementary combinations, that is, the categories of potential
mutual substitution, after which I will formulate some preliminary hypotheses:

Panaka is never combined with Yiparrka
Paljeri is never combined with Tjarurru, Milangka or Yiparrka
Burgulu is never combined with Milangka or Yiparrka

III.a First hypothesis: the Panaka-—Yiparrka identity

Panaka and Yiparrka are, following a purely combinatory approach, not categories of
mutual substitution. Indeed, the Ngaatjatjarra and the Ngaanyatjar-ra have each been divided
into two sub—groups, because both dialectal groups use a total of six terms. Nevertheless, we
are able to consider these two categories as “identical”, because Panaka and Yiparrka are in
an identical structural position in these two groups: the intermarrying category for both is
Tjarurru. Thus, when Panakn and Yiparrka appear together in the same system, they function
as synonyms, reinforcing the hypothesis that they are mutual substitutes for each other among
other groups and systems.

Indeed, for Bates cited in Elkin 1940:3312, Yrparrka replaces Panaka. This thesis seems
accurate for the Mandjindja, a neighbouring group of the Ngaatjatjarra and Ngaanyatjarra,
but in other cases this substitution is doubtful. A globally recognised identity between Pauaka
and Yrparrka is not possible in all situations without threatening the internal coherence—
and, above all, the transformation—of the system see discussion below, in relation to the
Pintupi.

We shall need to distinguish two kinds of identities between categories: the first will be
called relational identity, and the second, assimilatary identity. Panaka and Yiparrka may