Antecedent Priming in Sentences with Neutral Scrambling: Evidence from Dutch and German

In Focus


Latest Vacancies

A new post has arisen for a Teaching Fellow in Language Acquisition for the .within the Research Department of Linguistics. Follow this link for more information More>>


News and Events

For all the latest news and event details within the Faculty of Psychology and Language Sciences follow this link. More>>


Why Study here?

I chose London, not only due to the fact that it is a very cosmopolitan city, but also for its well known reputation in this area: More>>


Divisional subject pool

To access and sign up for the Divisional Subject Pool, follow this link.  More >>


Research Context

Previous linguistic research on NS has almost exclusively been informed by individual researchers' intuitive judgements and by theory-internal considerations, so that existing claims and hypotheses regarding the nature of NS are not currently based on a solid empirical footing. There is anecdotal evidence that NS does not exhibit scope-reconstruction, which – if correct – would support a base-generation analysis. The use of suitable experimental data-gathering techniques will provide us with new types of data to extend and improve the existing knowledge base and should allow us to empirically dissociate conflicting theoretical hypotheses about NS. 

There is psycholinguistic evidence from several experimental techniques (including EEG, fMRI and cross-modal priming (CMP)) that corroborates the linguistic analysis of wh-questions and other types of A’-dependencies as involving movement of an operator expression to the left clausal periphery. Moreover, the evidence from CMP – a technique that is well suited to detecting position-specific reactivation effects – indicates that the antecedent in such dependencies is linked to a trace (the Trace Reactivation Hypothesis (TRH); Love & Swinney 1996, Nakano, Felser & Clahsen 2002, Nicol & Swinney 1989, Swinney, Zurif & Nicol 1989, Zurif et al. 1993, 1995) rather than directly to a selecting verb (the Direct Association Hypothesis (DAH); Pickering & Barry 1991). There has been less focus on A-dependencies, the non-canonical placement of arguments found in passives, raising constructions and unaccusatives, although here too, evidence is emerging pointing towards movement (Osterhout & Swinney 1993, Friedmann et al. 2008, and Shetreet et al. 2010). Our research contributes to the experimental study of A-dependencies in the domain of scrambling.

Crucially, studies looking at short scrambling have not generally distinguished between NS and short A’-scrambling, which may well be responsible for the murky picture that emerges from these studies. A notable exception is Clahsen and Featherston's (1999) study of scrambling in German, which presented test sentences in contexts that strongly favour NS. Their first experiment used sentences with the structure in (3a), in which the verb has been moved to second position leaving a sentence-final trace (tV). As a result, the probe site for the scrambled direct object (tDO) is also sentence-final. A second experiment used particle verbs, so that the probe site preceded a stranded particle (Prt), as in (3b). 

(3) a. SUBJECT V DO IO (tDO) tV
b. SUBJECT V DO IO (tDO)  tV Prt

Remarkably, the first experiment yielded no evidence for reactivation of the direct object (DO) in sentence-final position, while the second experiment did find reactivation of DO just before Prt. One might speculate that these results indicate that the sentence processor accesses thematic information in the preposed verb. In (3a) the remainder of the sentence is compatible with this information, while in (3b) the indirect object (IO) can only be thematically licensed once Prt has been encountered. Thus, examples with the structural pattern in (3b) may require reanalysis of the verb's argument structure in a way that examples with the structure in (3a) do not. But crucially, neither experiment probed for reactivation of arguments other than DO.

In this project, we carry out two sets of experiments, one for Dutch and one for German, that aim to establish whether there is position-specific and argument-specific reactivation of scrambled arguments. The Dutch experiments address subquestion 1 and investigate structures involving NS across an adjunct, whereas the German experiments address subquestion 2 and aim to clarify the status of NS across an argument.