| |
Abstract:
The more frequently a word occurs, the easier it is to
recognize. This basic effect can be modified by a number of other
factors, e.g. sentence context. The goal of the positron emission
tomography study reported here was to investigate 1) which areas of
the brain are influenced by frequency and 2) how the activation
found in these areas is affected by the interaction of the
following factors: frequency, sentence vs. non-sentence context and
sentential complexity.
Methods: Subjects were presented with sentences containing high
frequency content words during four scans and sentences containing
low frequency content words in four other scans. The sentences
varied in syntactic complexity. Lists presented during four
additional scans were created by scrambling sentence lists
containing either low or high frequency content words to make
unconnected word lists. Thus effects of word frequency, sentence
vs. word list context and sentential complexity on regional blood
flow could be investigated, as well as interactions of these
factors.
Results: Both sentences and word lists containing low frequency
content words activated the left occipital cortex more than those
containing high frequency words. This difference suggests that
visual processing differs for low and high frequency words during
word recognition. This occipital activation was, however, sensitive
to the context in which the words occurred, with more difference
between the two levels of frequency in sentences than in word
lists. This result suggests that the early stages of visual word
processing are influenced by the sort of context in which words are
read.
There was also a main effect of sentences vs. word lists in the
anterior temporal lobe, replicating previous results (cf. Mazoyer
et al, 1993; Stowe et al, 1999; Tzourio et al, 1997). We have
previously suggested that this area is involved in encoding lexical
(semantic or syntactic) information into temporary storage, so that
it remains available/active for use later in the sentence or
discourse (Stowe et al, 1999). As predicted by this hypothesis,
this area is more activated by sentences containing low frequency
content words, which require additional effort to encode, than by
sentences containing high frequency words.
Mazoyer et al, 1993, Journal of Cognitive Neuroscience 5:
467-479.
Stowe et al, 1999, Psychophysiology 36: 786-801.
Tzourio et al, 1997, NeuroReport 9: 829-833.
|
