Each year the Conference on Computational Natural Language Learning (CoNLL) 1 features a shared task, in which participants train and test their systems on exactly the same data sets, in order to better compare systems. The tenth CoNLL (CoNLL-X) saw a shared task on Multilingual Dependency Parsing. In this paper, we describe how treebanks for 13 languages were converted into the same dependency format and how parsing performance was measured. We also give an overview of the parsing approaches that participants took and the results that they achieved. Finally, we try to draw general conclusions about multi-lingual parsing: What makes a particular language, treebank or annotation scheme easier or harder to parse and which phenomena are challenging for any dependency parser? Acknowledgement Many thanks to Amit Dubey and Yuval Krymolowski, the other two organizers of the shared task, for discussions, converting treebanks, writing software and helping with the papers. 2

We present a two-stage multilingual dependency parser and evaluate it on 13 diverse languages. The first stage is based on the unlabeled dependency parsing models described by McDonald and Pereira (2006) augmented with morphological features for a subset of the languages. The second stage takes the output from the first and labels all the edges in the dependency graph with appropriate syntactic categories using a globally trained sequence classifier over components of the graph. We report results on the CoNLL-X shared task (Buchholz et al., 2006) data sets and present an error analysis. 1

This paper provides an algorithmic framework for learning statistical models involving directed spanning trees, or equivalently non-projective dependency structures. We show how partition functions and marginals for directed spanning trees can be computed by an adaptation of Kirchhoff’s Matrix-Tree Theorem. To demonstrate an application of the method, we perform experiments which use the algorithm in training both log-linear and max-margin dependency parsers. The new training methods give improvements in accuracy over perceptron-trained models. 1

To facilitate future research in unsupervised induction of syntactic structure and to standardize best-practices, we propose a tagset that consists of twelve universal part-of-speech categories. In addition to the tagset, we develop a mapping from 25 different treebank tagsets to this universal set. As a result, when combined with the original treebank data, this universal tagset and mapping produce a dataset consisting of common parts-of-speech for 22 different languages. We highlight the use of this resource via three experiments, that (1) compare tagging accuracies across languages, (2) present an unsupervised grammar induction approach that does not use gold standard part-of-speech tags, and (3) use the universal tags to transfer dependency parsers between languages, achieving state-of-the-art results.

A strong inductive bias is essential in unsupervised grammar induction. We explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. Specifically, we investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In experiments with 12 languages, we achieve substantial gains over the standard expectation maximization (EM) baseline, with average improvement in attachment accuracy of 6.3%. Further, our method outperforms models based on a standard Bayesian sparsity-inducing prior by an average of 4.9%. On English in particular, we show that our approach improves on several other state-of-the-art techniques. 1

We describe our entry in the CoNLL-X shared task. The system consists of three phases: a probabilistic vine parser (Eisner and N. Smith, 2005) that produces unlabeled dependency trees, a probabilistic relation-labeling model, and a discriminative minimum risk reranker (D. Smith and Eisner, 2006). The system is designed for fast training and decoding and for high precision. We describe sources of crosslingual error and ways to ameliorate them. We then provide a detailed error analysis of parses produced for sentences in German (much training data) and Arabic (little training data). 1

This paper develops a general framework for machine learning based dependency parsing based on a pipeline approach, where a task is decomposed into several sequential stages. To overcome the error accumulation problem of pipeline models, we propose two natural principles for pipeline frameworks: (i) make local decisions as reliable as possible, and (ii) reduce the number of sequential decisions made. We develop an algorithm that provably satisfies these principles and show that the proposed principles support several algorithmic choices that improve the dependency parsing accuracy significantly. We present state of the art experimental results for English and several other languages. 1 1

A strong inductive bias is essential in unsupervised grammar induction. In this paper, we explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. We use part-of-speech (POS) tags to group dependencies by parent-child types and investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In experiments with 12 different languages, we achieve significant gains in directed accuracy over the standard expectation maximization (EM) baseline for 9 of the languages, with an average accuracy improvement of 6%. Further, we show that for 8 out of 12 languages, the new method outperforms models based on standard Bayesian sparsity-inducing parameter priors, with an average improvement of 4%. On English text in particular, we show that our approach improves performance over other state of the art techniques.

We describe an online learning dependency parser for the CoNLL-X Shared Task, based on the bottom-up projective algorithm of Eisner (2000). We experiment with a large feature set that models: the tokens involved in dependencies and their immediate context, the surfacetext distance between tokens, and the syntactic context dominated by each dependency. In experiments, the treatment of multilingual information was totally blind. 1

Probabilistic grammars offer great flexibility in modeling discrete sequential data like natural language text. Their symbolic component is amenable to inspection by humans, while their probabilistic component helps resolve ambiguity. They also permit the use of well-understood, generalpurpose learning algorithms. There has been an increased interest in using probabilistic grammars in the Bayesian setting. To date, most of the literature has focused on using a Dirichlet prior. The Dirichlet prior has several limitations, including that it cannot directly model covariance between the probabilistic grammar’s parameters. Yet, various grammar parameters are expected to be correlated because the elements in language they represent share linguistic properties. In this paper, we suggest an alternative to the Dirichlet prior, a family of logistic normal distributions. We derive an inference algorithm for this family of distributions and experiment with the task of dependency grammar induction, demonstrating performance improvements with our priors on a set of six treebanks in different natural languages. Our covariance framework permits soft parameter tying within grammars and across grammars for text in different languages, and we show empirical gains in a novel learning setting using bilingual, non-parallel data.