We apply the Stat-XFER statistical transfer machine translation framework to the task of translating from French and German into English. We introduce statistical methods within our framework that allow for the principled extraction of syntax-based transfer rules from parallel corpora given word alignments and constituency parses. Performance is evaluated on test sets from the 2007 WMT shared t...

It is well known that parsing accuracies drop significantly on out-of-domain data. What is less known is that some parsers suffer more from domain shifts than others. We show that dependency parsers have more difficulty parsing questions than constituency parsers. In particular, deterministic shift-reduce dependency parsers, which are of highest interest for practical applications because of th...

PARSEVAL, the default paradigm for evaluating constituency parsers, calculates parsing success (Precision/Recall) as a function of the number of matching labeled brackets across the test set. Nodes in constituency trees, however, are connected together to reflect important linguistic relations such as predicate-argument and direct-dominance relations between categories. In this paper, we presen...

We leverage innovative spatial modeling techniques and data on the precise geo-locations of more than 34,000 Constituency Development Fund (CDF) projects in Kenya to test whether MPs reward their supporters. We find only weak evidence that MPs channel projects disproportionately to areas inhabited by their political allies, once we control for other factors that affect where projects are placed...

I have argued that the assumption of incremental left-to-right syntactic structure-building can explain a number of otherwise puzzling contrasts in the results of syntactic constituency tests, such as coordination, binding, movement and deletion (Phillips, 1996, 2003). Briefly, non-root merger in a left-to-right derivation leads to systematic changes in constituency over the course of the deriv...