In a 2 x 2 between-groups study, 85 preservice education students observed
a science experiment concerning either buoyancy or the forces acting on a s
tationary object. Each student then wrote an Initial explanation of the phe
nomenon followed by a journal-style note, then a final explanation. For eac
h science experiment half the students received a list of strategy prompts
intended to facilitate learning through writing, and half wrote without the
se prompts. Forty-three percent of the "buoyancy" students and 14% of the "
forces" students increased the complexity of their explanations during the
writing interval. Strategy prompting did not increase explanatory gains. Te
xtual analysis showed that for the buoyancy problem, writing comparisons am
ong trials and explanations of individual trials correlated with explanator
y gains during the writing interval. For the forces problem, writing a conc
luding summary correlated negatively with explanatory gains. Qualitative an
alysis suggested that rhetorical structures (explanation, comparison, argum
entation, and summarization) contributed to learning during three phases of
building explanations: reviewing experimental trials, analyzing these tria
ls to identify causal variables, and generalizing these analyses to form ne
w explanations. These rhetorical structures stimulated, rather than structu
red, the construction of new knowledge and mapped onto the logical operatio
ns through which writers coordinated hypotheses and experimental trials in
a many-to-many, rather than a one-to-one,fashion.