ULTRA: Universal Grammar as a Universal Parser.

Area of Science:

• Linguistics
• Generative Grammar
• Computational Linguistics

Background:

• Traditional generative grammar views hierarchy and word order as dimensions of a single syntactic representation.
• Existing models often use process-neutral grammars and focus on Merge as a core syntactic operation.
• The concept of directionality in grammar and its connection to performance mechanisms are increasingly explored.

Purpose of the Study:

• To propose a novel model of universal grammar (UG) as a one-directional, universal parser.
• To reframe the relationship between word order (language-particular) and hierarchy (universal) as temporal sequences.
• To demonstrate how this model accounts for linearization, displacement, composition, and syntactic cycles.

Main Methods:

• Developed a universal parser model based on Knuth's stack-sorting algorithm.
• Analyzed word order in noun phrases as a key empirical domain for cross-linguistic generalizations.
• Investigated the concept of syntactic cycles as extended projections of lexical heads, linking to the notion of phases.

Main Results:

• The model unifies linearization, displacement, and composition into a single parsing process.
• It provides a novel, unambiguous source for labeled brackets, interpreted as parsing routines.
• The model explains 213-avoidance in word order and grounds syntactic cycles, dispensing with parameters.

Conclusions:

• The proposed one-directional parser model offers a new perspective on universal grammar, treating word order and hierarchy as temporal sequences.
• This architecture simplifies core syntactic processes and provides a unified account for specific word order constraints.
• The model's focus on noun phrases and cycles offers insights into cross-linguistic generalizations and the nature of recursion in language.