46
taxon either). Waddy gets around this problem by the wholesale use of ‘covert com-
plexes’, but this has the appearance of a solution to a problem that exists only in the
analyst’s mind. The discussion in Sect. 2.4.2 regarding the objectivity and consistency
of morphology -based scientifi c taxonomy also calls into question the signifi cance of
any observed similarities of folk and scientifi c taxa, with regard to monotypy.
The existence of classifi catory ‘ranks’ is key to Berlin ’s conception of the folk
genus. The generic rank of a folk classifi cation is meant to be the most salient to
speakers of that language, partly because linguistic consultants provide ‘generic’
terms when asked by someone (usually the ethnobiologist) to name a plant or ani-
mal. However, it is worth asking oneself whether an interview situation (formal or
informal), where someone from outside the speech community interviews a native
speaker about plant name s is really a natural speech situation that the latter would
encounter often in his/her lifetime. Actual naturalistic language data are rarely, if
ever, presented in ethnobiological studies, and it is therefore impossible to base
claims of salience on anything other than the investigator’s own subjective impres-
sions. The few exceptions include Baker [ 13 ], who argues that in the Australian
languages Ngalakan, Wubuy and Arrente, binomial compounds can be constructed
not to refer to specifi c level taxa, but only to describe individuals of a species (along
the lines of big dog, or black cat). The great majority of taxon names (even the
names of polytypic scientifi c genera) in these languages are unanalysable mono-
morphemic labels. McKnight [ 77 ], on the other hand, demonstrates that the com-
mon way of announcing an upcoming hunt for turtle or dugong in Lardil is ngada
kendabaliyathu , using the superordinate category kendabal ‘dugong/turtle’, instead
of the generic terms dilmirrur ‘dugong’ and bararun ‘turtle’. Here, the superordi-
nate term is clearly more salient in the context of a particular kind of announcement
(e.g. about a hunt from a canoe, using a spear).
Another piece of evidence that Berlin forwards to support the primacy of the
generic rank is the child language acquisition study by Stross [ 140 ], which shows
that Tzeltal children (this cross-sectional study involved children around the age
of four) learned the generic names of plants and animals before the specifi c and
varietal names ([ 9 ] p. 63). The results of the Stross study were verifi ed by a fol-
low-up study by Zarger and Stepp [ 143 ]. However, recent child development stud-
ies have shown that even pre-verbal infants are able to form global categories (i.e.
categories equivalent to ‘animal’ versus ‘furniture’) well before they are able to
form a basic- level category (i.e. ‘cat’ versus ‘ elephant ’) [ 144 ]. This fi nding has
been replicated in a longitudinal study by Pauen [ 145 ]. The results of these experi-
ments clearly show that the developmental trajectory of pre-verbal infants goes
moves from global to basic level categorisation , and not the other way round. If
Stross’ data, based on interviewing 4-year olds, is meant to demonstrate the pri-
macy of generic over specifi c and varietal taxa, then surely Quinn and Johnson’s
results on 2-month-olds would strongly suggest that higher-order taxa are more
salient than generic taxa.
Berlin ’s concept of rank is closely linked to his rules of ethnobiological nomen-
clature. Points 10 and 11 (see above) of his model require taxa of certain rank to
2 Ethnotaxonomies and Universals: Investigating some Key Assumptions