Processing gapped sentences

Frank Wijnen¹, and Edith Kaan²
Kaan@Duke.edu, frank.wijnen@let.uu.nl
1 Utrecht University, Utrecht Institute of Linguistics OTS
Trans 10, 3512 JK Utrecht, Netherlands
2 Duke University, Center for Cognitive Neuroscience
Box 90999, Durham NC 27708-0999, USA

In this study we explore the processing of an empty category anaphor not resulting from movement, viz. gapped verbs in sentence coordination, as in "Tina sent a letter to her father, and Iris ___ a bouquet to her mother". Processing studies of gapping may take on a special significance against the backdrop of deep linguistic controversies, the most dramatic of which concerns the question whether gapped verbs are syntactically represented or not. The questions we address in this study are (1) When does the sentence processor recognize a gapped verb? (2) Does recognition of the gap prompt a reconstructive process, and if so, when? One relevant study, Günther et al. (1993), reports self-paced reading data suggesting that gapped verbs are not reconstructed online. We ran a self-paced reading study (in Dutch), exploiting a plausibility manipulation illustrated in (1):

(1)

a.

Tina sent a letter to her father, and Iris a bouquet to her mother

b.

Tina wrote a letter to her father, and Iris a bouquet to her mother.

A difference in reading time between 1a and 1b on the word following the object NP in the second con-junct (a bouquet) indicated that features of the gapped verb were reactivated. Additionally, a comparison of gapped constructions to string-identical non-gapped control sentences did not yield any indication that reconstructing the verb information is costly. The results of an ERP study in English, which made use of the same plausibility manipulation, confirmed the SPR data. ERPs to the object noun in the second (gapped) conjunct show an N400 followed by a positivity (P600) if this word and the gapped verb yield an implausible combination (wrote ... a bouquet), suggesting that verb information is available and made use of at this position. The results of a cross-modal lexical decision experiment (in Dutch), finally, indicated that there is reactivation of the verb information at the very first position where the verb gap can be detected (given the verb second rule for Dutch main clauses), which indicates that syntax has a role to play in processing these constructions. Taken together, the results indicate that (1) information pertaining to the gapped verb is reactivated on-line. (2) This reactivation is supported by syntactic processing. (3) There is a minimal delay between gap detection and reactivation. (4) There are no indications that reactivation is costly.

References

Günther, U., Schade, U, Sichelschmidt, L. & Strohner, H. (1993). Elliptische Koordination. Strukturen und Prozesse lokaler Textkohärenz. Linguistische Berichte, 146, 312-342.