Christian J. Fiebach, Ina Bornkessel & Angela D. Friederici
Max Planck Institute for Cognitive Neuroscience, Leipzig
Approaches to (possible) individual differences in sentence processing have hitherto focused mainly on quantitative distinctions between the processing abilities of different groups of participants (e.g., Just & Carpenter, 1992; MacDonald, Just, & Carpenter, 1992). Thus, high span readers are typically classified as more "efficient" sentence processors than low span readers. However, research into non-linguistic processing differences between high and low span readers indicates that the distinction between the two groups may be qualitative, rather than quantitative, since low span readers generally show a greater susceptibility to the interference of irrelevant information (e.g., Chiappe, Hasher, & Siegel, 2000). The present ERP study examined whether the degree of interfering information also plays a role in determining (qualitative) individual processing differences in sentence comprehension. To this end, two groups of participants (high and low span readers) read ambiguous sentences such as (1), the competing readings in which should increase the degree of interference, as well as unambiguous controls.
| (1) | Klaus fragte sich, welche Sängerin am Sonntag nachmittag hinter der Kirche den/der Richter gesehen hat. |
| Klaus asked himself [which singer]AMB on Sunday afternoon behind the church [the judge]ACC/NOM seen has |
Whereas high span readers showed a sustained negativity with a left-anterior focus for ambiguous in comparison to unambiguous sentences throughout the ambiguous region, low span readers showed a sustained posterior positivity in the same contrast. Furthermore, only high span readers showed a P600 when sentences were disambiguated towards a dispreferred reading. These results are in line with a proposal advanced in Friederici, Steinhauer, Mecklinger, & Meyer (1998) that high span readers process ambiguous sentences more efficiently because they more effectively inhibit dispreferred readings. Thus, low span readers must invest more effort into the suppression of such readings (hence the sustained positivity), while the processing of high span readers is better described in terms of a continued activation of the preferred reading (cf. Fiebach, Schlesewsky & Friederici, 2001, for evidence in this regard from the processing of unambiguous sentences).