Focus (In)sensitivity

We define class A to include quantificational adverbs (e.g. "always") and class B to include "only" and "even". We firstly observe that researchers have systematically failed to apply any single theory in equal detail to both classes. Then, based on original data from English, German, Swedish, Dutch and Italian, including the extraction data below, we claim no current theory can account for both classes.

1. Mary knows which man(i) Jane said she (A) always/(B) only looked at t(i).
2. Mary knows which man Jane said was such that she (A) always/(B) *only looked at HIM.

While 1(A) has the reading in 2(A), 1(B) lacks the equivalent reading in 2(B). A similar contrast is observed in Swedish and Italian.

No current theory distinguishes between classes A and B, so no theory explains these contrasts. We show that similar patterns emerge on a range of heterogeneous data. We claim that only class B should be viewed as FSPs. In contrast, we claim that class A is made up of anaphoric expressions which are sensitive to discourse topic.