Mahurin Honors College Capstone Experience/Thesis Projects
Simulating the Machine Translation of Low-Resource Languages by Designing a Translator Between English and an Artificially Constructed Language
Natural language processing (NLP), or the use of computers to analyze natural language, is a field that relies heavily on syntax. It would seem intuitive that computers would thrive in this area due to their strict syntax requirements, but the syntax of natural languages leaves them unable to properly parse and generate sentences that seem normal to the average speaker. A subfield of NLP, machine translation, works mainly to computerize translation between different languages. Unfortunately, such translation is not without its weaknesses; language documentation is not created equal, and many low-resource languages—languages with relatively few kinds of documentation, most often written—are left with no way to effectively benefit from machine translation. As a step toward better translation processors for low-resource languages, this thesis examined the possibility of machine translation between high resource languages and low resource languages through an analysis of different machine learning techniques, and ultimately constructing a simple translator between English and an artificially constructed language using a context-free grammar (CFG).
Advisor(s) or Committee Chair
Trini Stickle, Ph.D.
Computational Linguistics | Computer Sciences | Linguistics
Snyder, Michaela, "Simulating the Machine Translation of Low-Resource Languages by Designing a Translator Between English and an Artificially Constructed Language" (2023). Mahurin Honors College Capstone Experience/Thesis Projects. Paper 1006.