Incrementality in naming and reading complex numerals: evidence from eyetracking

Abstract

Individuals speak incrementally when they interleave planning and articulation. Eyetracking, along with the measurement of speech onset latencies, can be used to gain more insight into the degree of incrementality adopted by speakers. In the current article, two eyetracking experiments are reported in which pairs of complex numerals were named (arabic format, Experiment 1) or read aloud (alphabetic format, Experiment 2) as house numbers and as clock times. We examined whether the degree of incrementality is differentially influenced by the production task (naming vs. reading) and mode (house numbers vs. clock time expressions), by comparing gaze durations and speech onset latencies. In both tasks and modes, dissociations were obtained between speech onset latencies (reflecting articulation) and gaze durations (reflecting planning), indicating incrementality. Furthermore, whereas none of the factors that determined gaze durations were reflected in the reading and naming latencies for the house numbers, the dissociation between gaze durations and response latencies for the clock times concerned mainly numeral length in both tasks. These results suggest that the degree of incrementality is influenced by the type of utterance (house number vs. clock time) rather than by task (reading vs. naming). The results highlight the importance of the utterance structure in determining the degree of incrementality.

Notes

This research was made possible by a PhD stipend from the Max Planck Society to the first author. We wish to thank Kay Bock, Marc Brysbaert, Marco Zorzi, the members of the Utterance Encoding Group at the Max Planck Institute for Psycholinguistics in Nijmegen, and one anonymous reviewer for helpful comments and discussions.

In all multilevel multiple regression analyses reported in this paper, the logarithm of response latencies and gaze durations was taken as the dependent variable, in order to obtain a better fit, by reducing the variability caused by outliers in the data. Importantly, a similar pattern of results is obtained (i.e., best fitting models with an identical set of factors) when taking the response latencies and gaze durations as the dependent variable instead.

It has been established that interword spaces benefit reading. However, a study by Inhoff, Radach, and Heller (2000), investigating the role of interword spaces in compound reading, has produced mixed results. They showed that interword spaces facilitate the access of the constituent word forms during the initial phase of compound reading, but hinder the subsequent specification of a conceptually unified compound meaning.