The purpose of the experiments described in this thesis is to
investigate the use of instability of spellings, made by seven year old
children, as the dependent variable in research into variables controlling
"learning to spell". A spelling is defined as unstable if, when a child
next writes the word, he uses different letters. "Instability" refers
to the total number of changed spellings in a number of words.
It is proposed that instability reflects the probability of different
spellings occurring. Skinner (1950 ) has suggested that measuring and
attempting to manipulate the probability of responding is the most efficient
method of studying "learning" .
Section One examines previous research into spelling development
with reference to the definitions and measurements of spelling used and
the theories which resulted from them. It is suggested that previous
definitions have invoked sub-skills and other explanations occurring
at different levels from the observed behaviour of spelling and that
variables controlling spelling will be more easily investigated if an
operational definition is used. Such a definition suggests that
instability should be measured.
Section Two describes experiments which establish that instability
has the necessary characteristics of a dependent variable. It occurs
in all children, is found in classroom work and can be reproduced in
experimental situations. However, it tends to decrease in repeated
testing. It was found that incorrect responses were more often unstable
than correct responses.
In Section Three experiments are described in which comment is made
on the correctness of spellings, in order to test the sensitivity of
instability. A three phase design is used in which comment was made
during the middle phase. It was found that comment affects instability
only if it is made after the child has read an incorrect spelling aloud.
It is proposed that measuring the instability of a spelling is a
dependent variable which will elicit information on variables controlling
the probability of spelling responses and thus circumvent the need to
use theories of events occurring at different levels to explain the
behaviour.