A substantial body of evidence has accrued in the literature for the idea that syntactic intervention effects can be alleviated by moving interveners ‘out of the way.’ A well-known example comes from Icelandic long distance agreement examples like (1): dative experiencers, such as "einhverri konu" in (1a), block long- distance plural agreement between the finite verb and a low nominative argument, but movement of the experiencer to the subject position seems to be enough to make this long-distance agreement possible.
(1) a.*Það virðast einhverri konu myndirnar vera ljótar. EXPL seem.3PL some woman.DAT paintings.the.NOM be ugly ‘It seems to some woman that the paintings are ugly.’
b. Einhverri konu virðast t myndirnar vera ljótar. some woman.DAT seem.3PL paintings.the.NOM be ugly ‘It seems to some woman that the paintings are ugly.’ (Sigurðsson and Holmberg 2008)
A popular treatment of these facts in the probe-goal framework of Chomsky (2000, 2001) is a two-step derivation where the dative is first moved to a position above T, and then T probes to find its goal, with the trace left by A-movement of the dative not counting as an intervener (see e.g. Holmberg & Hròarsdòttir 2004). Appealing as this explanation is, there are both theoretical and empirical problems. On the theoretical side, such derivations are countercyclic, and it is not clear why a trace would not count as an intervener if we are to adopt a copy-theoretic approach to movement (Chomsky 1995). On the empirical side, the idea that traces do not intervene at all is troubled by the fact that, in Icelandic, moving an intervener does not seem to suffice for allowing long-distance agreement for first or second person, as examples such as (2) show.
(2) *Henni líkaðir ðú. her.DAT like.2S you.S.NOM ‘She likes you.’ (Schütze 2003)
In this talk, I provide a new angle on the ameliorating effect of movement on intervention in Icelandic and beyond. The crucial ingredient of my proposal is ‘sideward movement’ (Hornstein 1999), where a subpart X of some root Y may be merged with some other element Z in a separate root. I show that this allows for derivations where only a subpart of a nominal (e.g. the NP) is merged into the intervening position, with the additional functional structure being added onto the NP in a sideward fashion. I show that not only does this solve the theoretical problems, it also allows us to explain a number of curious facts about the scope properties of moving (Nevins & Anand 2003) and gives us a way to explain the fact that some but not all features may be gotten ‘out of the way’ by movement.