الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Dependency grammar is based on word-word relations. It assumes that the structure of a sentence consists of lexical items that are attached to each other by binary asymmetrical relations, known as dependency relations. The main goal of this study is to illustrate the phase of designing and building a parsing system for Modern Standard Arabic depending on the theoretical framework of dependency grammar. The study also aims to conduct a broad survey of the linguistic framework of dependency grammar, its history, the different theories of dependency grammar finally the pros and cons of applying dependency grammar. This research also attempts to explore the previous efforts for building dependency parsing systems and treebanks. Moreover, it explores how these systems were built and evaluated. In order to accomplish the purpose of this study, an analytical descriptive approach has been adopted and a corpus of Arabic sentences collected from the ALTEC corpus has been gathered. The documents of ALTEC corpus were collected from the electronic articles published in the Arabic Wikipedia. The overall size of the ALTEC corpus is three million words. The objective was to analyze 321 sentences out of the whole corpus. These sentences were analyzed and parsed automatically through the Interactive ANalyzer IAN. IAN is a natural language analysis system built within the UNL framework. Many examples of the analyzed corpus is illustrated and explained in detail. The faced problems and limitations of this study are also discussed. A statistical evaluation is established that resulted in a level of adequacy of 97.9 % of the system