Investigating the plural formation in Maltese with naive discriminative learning

Abstract

This study applies Naive Discriminative Learning (NDL) to the plural formation system of Maltese. By analyzing a corpus of 2,369 Maltese nouns, the authors investigate the predictability of plural forms using NDL models. The study contrasts the performance of unigrams, bigrams, and trigrams as cues in predicting sound and broken plurals. Findings indicate that trigrams are most effective for broken plurals, while unigrams perform best for sound plurals, highlighting the importance of consonant-vowel structures in plural classification.