Fundamental
Similarities and Differences Between Study-Test and Anticipation
Item Presentation Procedures in the Learning of Linguistic Items:
Quantitative Tests of the Retention Interval Model Via Varied Parameter
Estimation Modes
C. Izawa |
The
retention interval model (Izawa, 1981b) was tested via 342 data points
based on 60,084 observations from 338 subjects in 6 experiments,
by using highly restrictive parameter estimation modes. First, all
parameters controlling the retention interval effects were estimated
solely from spaced practice data, and those controlling retrieval
effects for each intercycle interval variation, under the study-test
method only. In spite of this severe restriction of parameter estimations,
the model satisfactorily predicted data not only under the study-test
method, but also under the anticipation method. The same processes
of quantitative investigation were repeated, next, using parameters
estimated solely from data under the anticipation method. The model's
performances were also satisfactory for both data for its own and
for the study-test method. It is indeed remarkable that 335 data
points could be accounted for by using only 66 parameters estimated
exclusively from data under either method, whichever was selected.
Results suggest that anticipation and study-test methods seem to
differ little in terms of acquisition (S) and retrieval (T) processes,
but merely in terms of retention processes, inherently resulting
in more intervening-study-events (tS) and-test-events (tT) intervals
under the former than under the latter.
|