We have built a discriminative reordering model for the phrase-based machine translation system Moses, which is developed at the University of Edinburgh. The model is a maximum entropy classifier which incorporates a variety of feature functions to predict phrase orientation for machine translation. Two kinds of features reported in literature, namely lexical features and dependency path featur...

I propose a tree-rewriting grammar formalism, TAGLET, defined by the usual complementation operation and the simplest imaginable modification operation. TAGLET is context free and permits lexicalization of treebank parses (though TAGLET is only weakly equivalent in generative power to general CFGs), as well as straightforward exploration of rich hierarchical syntactic structures and detailed fe...

We describe an existing multilingual information extraction system that automatically detects event information on disasters, conflicts and health threats in near-real time from a continuous flow of on-line news articles. We illustrate a number of strategies for customizing the system to process social media texts such as Twitter messages, which are currently seen as a crucial source of informa...

Although state-of-the-art parsers for natural language are lexicalized, it was recently shown that an accurate unlexicalized parser for the Penn tree-bank can be simply read off a manually refined treebank. While lexicalized parsers often suffer from sparse data, manual mark-up is costly and largely based on individual linguistic intuition. Thus, across domains, languages, and tree-bank annotat...

Previous studies of spoken picture naming using event-related potentials (ERPs) have shown that speakers initiate lexical access within 200ms after stimulus onset. In the present study, we investigated the time course of lexical access in written, rather than spoken, word production. Chinese participants wrote target object names which varied in word frequency, and written naming times and ERPs...

In this paper, we describe a coreference solver based on the extensive use of lexical features and features extracted from dependency graphs of the sentences. The solver uses Soon et al. (2001)’s classical resolution algorithm based on a pairwise classification of the mentions. We applied this solver to the closed track of the CoNLL 2011 shared task (Pradhan et al., 2011). We carried out a syst...

Recent works on parsing have reported that the lexicalization does not have a serious role for parsing accuracy. Latent-annotation methods such as PCFG-LA are one of the most promising un-lexicalized approaches, and reached the-state-of-art performance. However, most works on latent annotation have investigated only PCFG formalism, without considering the Collins’ popular head-driven model, tho...

The objective of this work is to conduct a corpus-based diachronic investigation lexicalization patterns motion events as well attention manner in Arabic narrative writing. Motivated by the scarcity research on events, study aims identify main linguistic constructions used express novels and more importantly investigate whether shift salience has taken place last hundred years. draws Talmy’s fr...

PCFGs can grow exponentially as additional annotations are added to an initially simple base grammar. We present an approach where multiple annotations coexist, but in a factored manner that avoids this combinatorial explosion. Our method works with linguisticallymotivated annotations, induced latent structure, lexicalization, or any mix of the three. We use a structured expectation propagation...