Computational Dependency Theory

Dependency structures sit at a meeting point between linguistic theory and language technology. Computational Dependency Theory brings researchers from both sides of that conversation together, asking how these structures can be defined, annotated, connected to meaning, and used in natural language processing.

Edited by Kim Gerdes, Eva Hajičová, and Leo Wanner, this collection gathers revised papers from the 2011 International Conference on Dependency Linguistics in Barcelona. Its scope is broad but closely connected: formal accounts of dependency lead into corpus annotation and parsing, with attention to the linguistic choices that shape computational tools.

From a formal definition to practical analysis

The volume opens with work on defining dependency structures, then turns to the design of coherent surface-syntax functions for corpus annotation and methods for finding annotation errors in treebanks. These discussions make the underlying assumptions of linguistic data more visible—an important concern when annotated corpora are used to develop or evaluate language-processing systems.

Where syntax meets meaning

Several contributions investigate how dependency-based syntactic analysis can support semantic interpretation. Topics include deriving meaning from learner-corpus annotations, examining valency patterns in Chinese-language networks, and mapping semantic structures onto text through statistical language generation.

Grammars, corpora, and parsing

Other papers develop formal approaches to dependency and categorial grammars, including work on grammar construction and treebank development. The final group focuses on parsing: graph-based and transition-based methods, incremental parsing of partial utterances, comparisons between rule-based and data-driven approaches, and a hybrid method combining rules with corpus statistics.

For readers working across linguistics and NLP

This research-oriented collection will interest computational linguists, natural language processing researchers, formal linguists, and readers studying dependency grammar or syntactic annotation. Rather than treating parsing as an isolated engineering problem, the papers invite readers to consider how linguistic theory, annotation decisions, and computational methods inform one another.