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

42

III. The Spatial distribution of pairs of terms

The spatial distribution of individual terms confirms that understanding the diffusion of
section names is not only a linguistic problem but also one that is made considerably more
complex by the use of a greater number of terms than are structurally necessary. This means
that either another terminological system has encountered the original terminology from the
Pilbara area or fi'om the south—west if we go back even further in time, or that terms have
been locally “invented”.

The next step is compiling the frequency of joint appearances of two categories their
combination, in the same way as was done for the section names themselves when grouping
them into categories. Indeed, if two categories are never combined and they are localised or
dislocated categories, then it is possible, /i'om a generalpoint of view, that they are mutually
substitutable, that is. that at one stage in the diflusion, one term replaced the other. Table 3
summarises the number of combinations between categories.

Punmgu Karimara Panaka Paljeri Tjarurru Milangka Yiparrka Burgulu

Purungu 27

Karimarra 23 23

Panaka 19 19 19

Paljeri 1 3 13 1 1 13

Tjarurru 10 6 4 0 10

Milangka 7 3 3 0 4 7

Yiparrka 5 1 0 0 5 4 S

Burgulu 4 4 1 2 1 0 0 4

Table 3: Combinations of sections

Categories that never appear together—zero in Table 3r—wil1 be called “complementary
combinations”; they may theoretically be substitutable e.g. Tjarurru and Paljeri. Categories
for which the number of combinations is close to the total number of their appearances will be
called “impossible substitutions”. Panaka is, for example, combined with Punmgu 19 times
out of 19 appearances. It is, therefore, not possible for Panaka to replace Punmgu, or vice
versa. Let us discuss some of these relations.

° Punmgu is used by all groups and combined with all other categories. Punmgu has,
therefore, never been replaced by another category, and may be considered as the
core of the four-section system terminology.

' Karimarra is combined with all other categories, and is used among all groups” except
for Ooldea 2 and the Mandjindja. Karimarm is an inherent category of the four-

26 Karimarra is not used among one of each subgroup of the Ngaatjatjarra and the Ngaanyatjarra. Yet because these
subgroups are a methodological artefact see below, since both Ngaatjatjarra and Ngaanyatjarra groups possess a
four-section system using six tenns instead of four, Kari.marra must be considered as being used by both of these
g|'°“P5-