Sentence context modifies compound word recognition: Evidence from eye movements

Abstract

Two eye movement experiments are reported that examine the influence of sentence context on morphological processing. English compound words which vary in beginning lexeme frequency (Experiment 1) and ending lexeme frequency (Experiment 2) were embedded into sentence contexts that were either predictive of the compound word or were neutral with respect to the compound. A predictable sentence context reduced the effect of beginning lexeme frequency on first fixation and single fixation durations. However, sentence context did not modify effects of beginning and ending lexeme frequency in later fixation measures. These results further support the theoretical position that morphology plays a role at multiple levels within readers' mental lexicons. In addition, these results suggest that access to early morpho-orthographic processes can be influenced by sentence context, a finding that suggests an interactive relationship between sentence context and word recognition.

Keywords:

• Compound words
• Eye movements
• Morphology
• Reading

Acknowledgements

The experimental data reported in this paper were collected in partial fulfilment of a PhD dissertation conducted by BJJ at the University of Massachusetts, Amherst. The research was supported by a predoctoral fellowship from the National Institute of Mental Health (MH16745). I would like to thank my dissertation committee (Keith Rayner, Charles Clifton, Jr., Alexander Pollatsek, Marvin Daehler, and Donald Fisher) for their helpful comments on this research. I would also like to Matthew Vitiello for help creating the stimuli, as well as Bernhard Angele and Klinton Bicknell for statistical training. Finally, I would like to thank Victor Kuperman and an anonymous reviewer for helpful comments. Portions of these data were presented at the 47th annual meeting of the Psychonomic Society.

Notes

¹In this paper, the term “compound words” refers to words composed of two lexemes that are spatially unified. See Juhasz, Inhoff, and Rayner (2005) for research examining the role of spatial layout in English compound word processing.

²While average semantic transparency was controlled in the two experiments, compounds employed covered a range of transparencies. Past eye movement studies with Finnish (Pollatsek & Hyönä, 2005) and English (Juhasz, 2007) have demonstrated similar lexeme frequency effects for semantically transparent and opaque compounds in the eye movement measures explored in the present study. Thus, both types of compounds are decomposed initially during normal reading.

³As noted in the introduction, recent studies have implicated a role for morphological family size of the lexemes within compound words on eye movements that is separate from that of lexeme frequency, although the two are correlated with each other (see Juhasz & Berkowitz, 2011; Kuperman et al., 2008, 2009). In the present study, positional morphological family size was allowed to covary naturally with lexeme frequency.

⁴In addition to the compound words, 20 monomorphemic words were also embedded in predictable and neutral sentences to reduce the chance that readers would develop decomposition strategies. These words were matched to the compound words in both experiments respectively in terms of length, frequency, and familiarity. They showed a standard predictability effect on fixation duration measures.

⁵Additional measures were also analysed for both experiments. For all analyses, a maximal random effects structure was used except for first fixation of multiple in Experiment 2 which did not converge. In this instance, only random intercepts were used. Initial landing position and first fixations of multiple did not yield any statistically significant results (all t-values < ∣2∣). In Experiment 1, regressions out of the target word were less likely in a predictive sentence context, b =−0.67, SE=0.26, z =−2.56, p =.011. No other effects in the regression out measure reached significance in Experiments 1 or 2 (ps>.15).

⁶One item from Experiment 2, cheeseburger, which was not in the English Lexicon Project (Balota et al., 2007) database, did not contribute to the bigram frequency analysis.