12
Cig
Lockhead, 1980), 74% of the "clusters" contained only one or
two items.
However, the major problem with the analysis procedures
usea to date is that they value differences between
exemplars within a recall list. They have had the objective
of identifying semantic associative substructure within a
recall sequence rather than semantic intension criteria for
being recalled in the. first place. Such analyses ignore an
important semantic concern. They detect factors that vary
within the larger semantic field and that must, therefore,
not be critical to that field. For example, Coltheart &
Evans (1981) found that such factors as sea bird vs. land
ird and predator versus non-predator accounted for large
proportions of the variance in the free recall lists of
birds. This means that in the free recall of examples of
birds, subjects listed both sea birds and land birds,
predators and non-predators. This in turn means that those
two factors were not critically relevant in deciding whether
or not something should be listed as a bird. Thus, although
they did demonstrate that subjects diftered in the
substructure of their free recall, they did not explain how
those differences necessarily and reliably relates to
semantic intensional criteria.
Rather than searching for differences among the