Robin L. Hill & Roger P.G. van Gompel
University of Dundee
We employed sentences such as (1) in an eye-movement experiment to investigate a number of hypotheses about the effects of lexical frequency and sentence wrap-up on eye-movement behaviour.
(1a) The manager spoke to the secretary. A few days ago there had been a terrible misunderstanding.
(1b) The manager spoke to the ballerina. A few days ago there had been a terrible misunderstanding.
(1c) The manager spoke to the secretary a few days ago. There had been a terrible misunderstanding.
(1d) The manager spoke to the ballerina a few days ago. There had been a terrible misunderstanding.
These sentences contained a high frequent ("secretary") or low frequent word ("ballerina"), which occurred either at the end of the sentence or a few words before the end.
The first claim that we tested was whether sentences function as processing units (e.g., Fodor et al., 1974), such that readers complete processing a sentence before they move on to the next one. If this hypothesis is correct, we may find spill-over effects of frequency in the region "a few" when this region is part of the same sentence as the critical word "secretary/ballerina" [(1c) and (1d)], but not when this region is the beginning of the next sentence [(1a) and (1b)]. Alternatively, if readers continue processing the words in the previous sentence, spill-over effects should also be observed when "a few" is the beginning of the next sentence.
The second hypothesis that we explored was whether sentence wrap- up affects the size of the frequency effect. It has been shown (Mitchell & Green, 1978; Rayner et al., 2000) that readers engage in additional processing when they reach the end of a sentence. If the additional processing due to sentence wrap-up affects the depth of lexical processing, we expect a larger frequency effect when the critical word is at the end of the sentence than when it is not. Finally, we investigated whether lexical frequency and the presence of a full stop have an effect on landing positions at the following region.
Our results showed effects of frequency and position of the full stop on the critical word "secretary/ballerina" in first-pass times, but no interaction between these factors. This indicates that the frequency effect is unaffected by the additional processing that is due to sentence wrap-up. We observed no effects of first-pass reading times in the spill- over region. However, we did find effects of frequency and full stop in regression path times (first-pass including rereading) and an interaction between these factors which indicated that the spill-over frequency effect was larger when there was no intervening full stop than when an intervening full stop was present. We conclude that readers continued processing the critical word in the spill-over region when both were part of the same sentence. In contrast, if the spill-over region was the start of the next sentence, readers did not continue processing the preceding word to the same extent. Finally, the results showed an effect of full stop on the landing position at "a few", but no effect of frequency. This is consistent with the view that visual and lexical factors have different effects on eye movements (e.g., Rayner, 1998).