Jenkins, Stein, and Wysoski (1984) had children read unfamiliar words under relatively
natural reading conditions (emphasis was not placed on vocabulary learning). The main
independent variable was the number of encounters students had with the unfamiliar
word. Brief passages were constructed, each containing a target word whose meaning was
not likely to be known by elementary aged students (e.g., altercation, incarcerate). Each
passage also contained a synonym for, or words which strongly cued the meaning of the
target word. Eighteen low-frequency target words were divided into three sets of six,
which were counterbalanced across conditions. Ten such passages were constructed for
each word. Under the different exposure conditions, students read either two, six, or ten
passages for each of the six target words. Students in the 10-exposure condition read six
passages (one for each target word) per day over 10 days. Students in the six-exposure
condition read six passages spread over 10 days, while those in the two-exposure
condition read six passages on days 1 and 10. Thus, the distribution of passages for the
three-exposure conditions controlled for recency effects prior to post-testing.
Various vocabulary measures were created. Two were definitional tests, one requiring
students to choose a correct definition or synonym from several alternatives, the other
requiring students to apply a suitable synonym or definition for a target word which
appeared in a sentence with minimal context clues. Two other measures involved word
usage. One was a sentence anomaly test which required students to state whether a
sentence made sense. One sensible sentence and one nonsense sentence were created for
each word. The other usage measure was a sentence completion test. For each target word
a sentence stem was created which included the target word followed by four choices, only
one of which was correct. In addition, to obtain zero-exposure, in the control group
students were also tested on non-exposed words.
They found that children can acquire word meanings from context. For this to happen,
however, the words must be encountered several times, even when each encounter is
accompanied by strong context clues. Ability differences again appeared, with higher-
ability readers learning more word meanings than lower-ability students.
Moreover, Nagy et. al. (1987) assumed that the evidence of learning from context was not
captured by most studies, because such learning proceeds in small increments that the
61