ROMJIST Volume 29, No. 4, 2026, pp. 299-312, DOI: 10.59277/ROMJIST.2026.4.02
Jonathan SANDE, Tseren-Onolt ISHDORJ Modeling Mongolian with Formal Language
ABSTRACT: This paper seeks to model the Mongolian language using formal language theory, particularly employing context-free grammar (CFG), filling a gap for sentence-level analysis that has been missing for Mongolian natural language processing. Mongolian is an agglutinative language where nouns and verbs can take many forms by concatenating suffixes to a root. This makes it challenging to create a CFG, which is usually whole-word based. A generalized CFG for simple declarative sentences covering the eight noun cases of Mongolian was constructed. Roots and suffixes are handled as separate terminals. Other context-sensitive aspects of Mongolian, like vowel harmony and spelling rules are left for post-processing. A pushdown automaton (PDA) corresponding to the CFG for Mongolian was designed to recognize whether the generated sentence structure is correct or not. This lays the foundation and creates a computational baseline for more sophisticated formal language (FL) representations of Mongolian in the future, such as mildly context-sensitive (MCS) grammars. Finally, a software platform was developed to implement the Mongolian simple declarative sentence generation and recognition by CFG and PDA, respectively. The CFG in the platform, modeled after the generalized Mongolian CFG in the paper, generates 8190 unique sentences. This is 1.7% of all possible permutations of input words in sentences up to four words long, which the PDA validated.KEYWORDS: Context-free grammar; formal language; Mongolian; natural language processing; pushdown automataRead full text (pdf)