Processing contextual and lexical cues to focus: Evidence from eye movements in reading

Abstract

Three eye movement experiments investigated the interaction between contextual and lexical focus cues during reading. Context was used to focus on either the indirect or direct object of a double object construction, which was followed by a remnant continuation that formed either a congruous or incongruous contrast with the contextually focused object. Experiment 1 demonstrated that remnants were more difficult to process when incongruous with the contextually focused constituent, indicating that context was effective in specifying focus. Experiments 2 and 3 investigated the interaction between context and lexical focus arising from the particle only which specifies focus on the subsequent adjacent element. When only preceded both objects (Experiment 2), the conflict between lexical and contextual focus cues disrupted processing of the remnant element and was resolved in favour of the contextually focused element. However, when only was placed between both objects (Experiment 3), cue-conflict disrupted processing earlier in the sentence but did not appear to be fully resolved during on-line sentence processing. These findings reveal that the interplay between contextual and lexical cues to focus is important for establishing focus structure during on-line sentence processing.

Keywords:

• Focus particles
• Discourse processing
• Sentence processing
• Eye movements while reading

Acknowledgments

We wish to thank the anonymous reviewers for helpful comments on previous versions of the manuscript. AS was supported by the German Research Foundation (DFG), within the Collaborative Research Centre SFB 632 ‘Information structure’ (Project C3). KP was supported by the British Academy (Grant SG40210).

Notes

¹Available from (April 2010): http://www.psych.umass.edu/eyelab/software/

²Additional ANOVA analyses conducted for the log-transformed data did not lead to results that differed from the results of the LME models.

³ANOVA-analyses on the log-transformed reading times revealed a similar pattern of results as LME models.

⁴Traditional ANOVA analyses produced similar results as the LME models although the ANOVAs did not show all effects of the LME models in the post-remnant region.

⁵Statistics for Experiment 1: Interrogative sentences (IO-context vs. DO-context, 3,257 ms vs. 3,099 ms; b=170, SE=76, t=2.24); Subject + Verb region (IO-context vs. DO-context: 560 msec vs. 547 msec; b=11, SE=20, t=0.55)

⁶Interrogative sentences (IO-context vs. DO-context, Experiment 2: 2,856 ms vs. 2,773 ms, b=70, SE=58, t= − 1.21; Experiment 3: 2,845 vs. 2,840 ms, b=17, SE=54, t=0.32, id slope adjustment for Remnant Congruency and Context); Subject + Verb region (IO-context vs. DO-context, Experiment 2: 589 vs. 607 ms; b=16, SE=17, t=0.95; Experiment 3: 572 vs. 580 ms, b= − 8, SE=15, t = −0.58)