Parsing Akkadian Verbs with Prolog

Abstract: This paper describes a parsing/generation system for finite verbal forms in Akkadian, with the possible addition of suffixes, implemented in Prolog. The work described provides the framework and engine to interpret the D, N, and G stems along with accusative, dative and ventive endings.

• The paper introduces a Prolog-based parser that leverages nearly 200 DCG rules to parse and generate intricate Akkadian verb forms.
• It employs innovative vowel wildcard methods to address challenges in vowel length determination and morphological complexity.
• The system integrates with a Java-based GUI, providing a practical tool for researchers studying ancient Akkadian texts.

Overview of "Parsing Akkadian Verbs with Prolog"

Parsing Akkadian Verbs with Prolog presents a system for parsing and generating finite verbal forms in Akkadian, implemented using the Prolog programming language. It focuses on handling complex morphological structures, particularly the D, N, and G stems, alongside direct-object, indirect-object, and ventive suffixes. The approach utilizes Definite Clause Grammar (DCG) rules extending regular context-free grammars, allowing for the parsing and generation of these verbal forms.

Core Contributions

The paper's primary contributions lie in the design and implementation of a parser capable of interpreting Akkadian verbs through several innovative techniques:

1. Prolog-Based Parsing and Generation:
   • Utilizes Prolog's backtracking to test potential finite forms, ensuring compatibility with strong roots and a subset of weak roots. The system processes forms using a Prolog Normal Form (PlNF), either input manually or generated automatically.
2. Handling of Vowels:
   • The paper introduces methods to manage the complexity of Akkadian vowel representation. It employs vowel wildcards to mitigate the recursive issue of vowel length determination, facilitating the parsing of normalized forms.
3. Pattern Recognition and Rule-Based Parsing:
   • The parser encompasses approximately 200 DCG rules to successfully parse various verb forms, providing recognition across different tenses and modalities such as preterite, durative, perfect, imperative, precative, and vetitive.

Implementation Details

The system is designed for integration with a Java-based GUI, operating through a web server. Though the parser’s core is in Prolog, it employs Java wrappers for interaction with the web interface, thereby enhancing accessibility:

• Low-Level Rule Utilization:
  • DCG rules are directionally agnostic, enabling both parsing and generation with minimal modification. The system can generate multiple forms due to variability in theme vowels, supporting exploration through a dictionary.
• User Interface Considerations:
  • A Unix shell script facilitates communication between the web form inputs and the Java-parsed Prolog interpretation, while addressing performance overheads by utilizing a backend processor to manage computationally intensive tasks.

Implications and Future Directions

The paper indicates practical and theoretical implications for both computational linguistics and Akkadian studies. From a practical standpoint, the parser provides a tool for researchers and students to analyze Akkadian texts, potentially enriching our understanding of ancient languages. Theoretically, it offers insights into the application of DCG and logic programming for complex language parsing tasks.

Future research directions may focus on expanding the system to accommodate more intricate verbal forms and additional derived stems. Enhancements to the user interface to incorporate wildcard functionalities and assimilation rules could significantly improve usability and functional completeness.

Overall, the paper presents a substantive advancement in computationally parsing ancient languages, offering a framework adaptable for similar linguistic challenges. As the system evolves, it is poised to contribute further to both the fields of computational linguistics and historical linguistics.