measures used in many studies have failed to capture. To examine this assumption they
employed a variety of measures designed to capture partial knowledge. The researchers
had seventh and eighth graders read passages containing target words and then respond
to three levels of multiple choice tests. The easiest level required only that the reader have
some very general knowledge of a word (knowledge of its part of speech or its general
category) which were sufficient to get the item correct. An interview task was also used
in which readers were asked to give definitional information. The study yielded small but
robust effects for learning from context.
Nagy et. al. (1987) point to the issue of measurement sensitivity in discussing their
finding of effects of single contextual exposures in contrast to Jenkins et al.’s (1984)
conclusion that at least more than two exposures were needed. Only the most difficult
level of Nagy et al.’s range of multiple choice items was comparable to the difficulty level
of the multiple choice test used by Jenkins et. al. Furthermore, although both groups of
researchers asked students to produce definitions, the scoring criteria were much more
moderate for Nagy et. al.’s interview responses.
In later studies, Nagy and Herman considered other text and reader characteristics that
could affect learning from context. Herman et. al. (1987) found that learning from context
was facilitated by conceptually explicit text and higher reading ability. Nagy et. al. (1987)
found that the conceptual difficulty of a word or of the text, diminished the learning of the
word. But they failed to find an effect for reading ability, which is a departure from
findings about the relationship between reading ability and learning from context reached
by many previous studies.
Dickinson (1984) studied the effect of varying the exposure context of a new word. Using
children aged 4 and 11 years, he presented them with one novel word in either a
conversational context, a story context or an explicit definition. The effect of presentation
mode varied with age: 11-year-olds were better at word recognition from the definitional
context than from the story context, whereas the 4-year-olds Ieamt little from either
context. The older children were excellent on usage of the words they had heard in stories,
therefore they must have formed representations of these words. Some learning was
taking place from hearing a single presentation of a novel word, in all three contexts.
62