231contexts in which certain ‘colour’ terms are used can reveal that perceptual features
of objects other than hue might be the referents of such terms. Studies of folk tax-
onomy can also benefi t greatly from the inclusion of naturalistic speech data, as
illustrated by McKnight’s [ 77 ] example of a superordinate taxon being more salient
than a ‘ generic ’ level taxon in the context of a Lardil speaker announcing his intention
to go hunting. Such culturally relevant, everyday contexts clearly have the potential to
motivate utterances that are qualitatively very different from those elicited through
interviews. A worrying example of analytical exuberance based on very limited
empirical grounds is the ‘taxonomy’ of ancestral beings in the Australian language
Dalabon, constructed by Maddock [ 228 ], as a means of explaining the ‘anomalous’
classifi cation of the emu. Maddock’s taxonomy contains two key nodes which, in
Dalabon, have no name; there are also no additional data present in the paper to sup-
port the four-tiered structure of the taxonomy, and it appears that the entire schema
was more an attempt “ to account as economically as possible for the ordering of spe-
cies ” (p. 102) than an accurate description of Dalabon speakers’ mental representation
of the categorisation of ancestral beings. Wierzbicka [ 229 ] sums up the problem of
analytical overenthusiasm with characteristic precision:
To assume that people in all cultures have the concept of ‘sadness’ even if they have no
word for it is like assuming that people in all cultures have a concept of ‘marmalade’ and
moreover, that this concept is somehow more relevant to them than the concept of ‘plum
jam’, even if they happen to have a word for the latter but not the former. (p. 9)
The concept of a cognitive map as a mental representation of spatial relations—
one that is either analogous to, or even like, a cartographic map—can also be chal-
lenged when one considers empirical data consisting of people’s generalisations
about the information that is supposedly encoded in such maps. Baird et al. [ 230 ]
have demonstrated mathematically that certain methodologies used to elicit cogni-
tive maps signifi cantly ‘transform’ subjects’ mental representations in specifi c
ways; as a result, the authors conclude that “ cognitive science must take seriously
the realization that method and substantive theory are inseparable ” (p. 213).
Additionally, several empirical studies, using techniques such as distance and direc-
tion judgements, map recognition and map reconstruction, have shown that people
appear to entirely reorganize spatial information from the external environment
when producing mental representations (reviewed in [ 231 ]). Tversky proposes a
‘constructionist’ view of cognitive maps, a view where “ representations of the
visual world are constructed, and...systematic errors may be introduced in the con-
struction of representations as well as in retrieval of information from them ”
(p. 135). She summarises her stance on cognitive maps as follows:
As we navigate an environment, or make inferences from memory of one, we draw on
information from many different sources, from particular episodes in the environment and
schematic knowledge of the environment, from verbal descriptions and visual experience,
from information specifi c to the environment, and from general information about that kind
of environment... When all that information is put together, it does not necessarily form a
coherent picture, something that could be drawn on paper or modelled in three dimensions.
On the contrary, the different bits and pieces may very well confl ict with each other, some-
thing that would not be evident without an attempt to put them together. (p. 137)8.3 Fact or Fiction: TEK from the Analyst’s Point of View