m9 
— 
constraint paradigms of free recall have been developed. 
Constrained free recall is a learning task in which the 
subject is presented a finite set of items to be remembered, 
typically to test the subject's ability to generate or to 
maintain semantic organization among the items (Winters & 
Brzoska, 1976; Thompson, 1978; Orasanu, Lee, & Scribner, 
1979; Thomas & Bolton, 1979). Dunlap & Dunlap (1979) added 
another constraint by having the subjects generate the items 
to be learned. 
Various procedures have been described for analyzing 
the semantics of free recall data. These have included: 1) 
subject generation of, and scaling on, dimensions 
descriptive of the recalled exemplars, with subsequent: 
principle components analysis (Coltheart and Evans, 1981), 
2) multidimensional scaling of the distance between 
exemplars. in the recall sequence (Buschke, 1977; Rubin & 
Olson, 1980), and 3) hierarchical cluster analyels 
(Friendly, 1977). Some studies nave proposed making use of 
the temporal clustering information evident in free recall 
output (Friendly, 1977; Gruenewald & Lockhead, 1980). But 
unresolved nethodolonloal issues of deciding when a word 
occurs, what constitutes a temporal cluster, and which 
measuring techniques are appropriate make use of the 
temporal clustering information questionable, at least for 
the meantime. For example, in a recent study (Gruenewald &