Parallel allocation of visual attention during word recognition

C. Ucelli, J. Pynte

CREPCO, Université de Provence 24 av. R. Schuman, F-13621 Aix en Provence, France (e-mail:pynte@univ-aix.fr)

The purpose of the two experiments reported here was to provide further evidence of lexical influence on eye-movement control during the reading of compound words and to comment on possible mechanisms by which lexical access processes influence early eye movements components. The question of a possible parallel visual attention allocation on the words is also asked.

Eye-movements were recorded during the reading of compound words which were displayed with or without the hyphen (e.g.: "rond point" ; "bloc-note"). Specific inspection patterns where found in the no-hyphen condition where two perceptual units (separated by a blank) must be grouped together in order to access the compound-word lexical entry.

In order to assess the potential influence of pre-established inspection strategies, the composition of the experimental list was varied. In the "single-word" condition, filler items were long (seven character) words, while in the "two-word" condition, there were pairs of short words (three to four characters each) This manipulation did influence the oculomotor behaviour (interaction with the hyphen vs. no-hyphen factor), suggesting that eye movement control in reading is submitted to high-level strategic effects.

The aim of the second experiment was to determine whether a pair of words like "ami aux" is likely to activate the high frequency word "amicaux", whose first three and last three letters correspond to the first and second word in the pair, respectively. Such "pseudo-compound" stimuli where compared to sequences such as "but vin" for which no single lexical entry can be found.

Again, inspection strategies were found to depend on processes developed at the lexical level. A parallel allocation of visual attention on the two words would be possible.