Comp-trace effects: a comparative and psycholinguistic approach
Final Report Abstract
This project focused on the difference in sensitivity to the COMP-trace effect between English, German and Dutch. In English, COMP-trace effects lead to absolute ungrammaticality, but in Dutch, such structures are fully acceptable, whereas their acceptability in German is degraded, but not completely unacceptable. Since the COMP-trace effect has been presumed to be universal, and the three languages are closely related, this is an issue that needs to be resolved. Finding out the source of this difference is important in understanding the true nature of the COMP-trace effect. One of the central hypotheses within the project was that the strength of the COMP-trace effect is dependent on the relative ease with which subject and object gaps can be distinguished from each other. In English, this is through word order, a very salient cue, but German and Dutch rely on morphological cues, which are richer in German than in Dutch. For German, our experiments showed that the strength of the COMP-trace effect is watered down if not both subject and object are unambiguously case-marked. The absence of unambiguous case morphology on an overt argument in the embedded clause leads to it being interpreted as the local subject, so that the subject gap is either ignored or not posited. In this case, no COMP-trace effect is found in grammaticality judgments either. However, no evidence was found for a direct relation between comprehension and grammaticality judgments. Absence of morphological disambiguation cues in the embedded clause simply seems to make a COMP-trace violation less noticeable. For Dutch, a grammaticality judgment task showed that COMP-trace effects could be elicited in temporary ambiguous structures that were disambiguated towards a subject-gap reading through verbal agreement. This is the first grammaticality judgment study that provided clear evidence for a COMP-trace effect in Dutch and suggests it can be experimentally induced using disambiguating morphological cues. For English, the project investigated the strategy that it uses to circumvent a COMP-trace effect, which consists of not using the (overt) complementizer. In a novel analysis developed within the project, it is proposed that there is no subject gap in the embedded clause, but a null subject pronoun that undergoes agreement with a structurally higher scope marker. An important conclusion that is drawn from the results is that the COMP-trace effect is best described as a processing-based constraint, which is grammaticalized in English, but not in German in Dutch.
Publications
-
COMP-trace effects in German: the role of processing. Nordlyd, 44(1), 117-132.
Schippers, Ankelien; Vogelzang, Margreet & Öwerdieck, David
