Morphological Segmentation During Silent Reading

This study tested two hypotheses about the properties of morphological segmentation: (a) that it applies on phonemic representations, and (b) that it outputs affixal information that is taken up and used at representational levels higher than the lexical one. In two experiments, participants\u27 eye-movements were monitored while they silently read sentences where the monomorphemic members (guest; bale) of monomorphemic-polymorphemic (MP) pairs of heterographic homophones (guest-guessed) and of monomorphemic-monomorphemic (MM) pairs of heterographic homophones (bale-bail) were embedded. The results of the first experiment provided evidence that morphological segmentation applies on phonemic representations in the absence of orthographic cues, as the MP homophones (guest) induced a processing cost in First Fixation in the subset of the data where they were preceded by an adjective-dominant modifier. A cost emerged clearly in First Fixation and Gaze Duration in Experiment 2, as well, where the homophones were preceded by balanced noun-adjective modifiers, and sentence context was tightly controlled. The cost in the MP condition in Quasi-First Pass Time and Total Time in the adjective-dominant modifier subset of the data in Experiment 1 suggested that the information activated via morphological segmentation remains active and is used at levels of representation higher than the lexical one. Experiment 2 tested this hypothesis directly. Longer rereading times were found for the MP homophones again. In addition, readers regressed, from later regions of text, less often to the adverb placed between the homophone/control and the matrix clause verb, in the MP condition. This was interpreted as indicating ease of integration of the adverb with the verb phrase it modifies, via the affix information made available by morphological segmentation