Cyclic Linearization or Pied Piping?

In celebration of the fact that I've just finished a major essay (only hours before going away on a week's skiing holiday), I thought I'd post an almost indecipherable blog about the issues covered in the essay, written for a Generative Grammar course (Intermediate B) as part of my linguistics BA at UCL.

The essay looks at Object Shift in Scandinavian languages, asking why it's possible to say 'I have not kissed her', 'I kissed her not' and sometimes 'I have her not kissed' but never *'I kissed her not'.

There have been a few ideas, advanced by Chomsky, Bobaljik and Holmberg, but the essay looks at two more recent ideas, from Fox and Pesetsky (2003) and Nilsen (forthcoming - he wrote a quick summary of his new paper exclusively for us!).

F&P's story (for some reason we refer to theories as stories, which says a lot) suggest that languages have Spell-Out domains at various levels, and that linear word order must remain consistent between these levels. In order to capture differences between English and Scandinavian, they stipulate that the English lower domain contains the external argument (subject) whereas Scandinavian languages do not.

Nilsen adds the ungrammatical Swedish sentence *'Therefore liked it evidently not John' into the mix, showing that F&P cannot account for this. He then goes on to give a completely different story, based upon a combination of Remnant Movement (where a word moves out of a phrase, and then the phrase moves separately) and a Pied Piping constraint (where a word being extracted from a phrase must pied pipe the constituent immediately dominating it).

This mix of constraints and syntactic phenomena does give account for all the data, but it isn't all that neat. And they're both fairly stipulative.

Welcome to generative grammar.