Numeral classifiers are additional grammatical elements within a numeral phrase (NumP) which refer to the salient semantic properties of the counted noun (Gil 2013). Numeral classifier systems are...Show moreNumeral classifiers are additional grammatical elements within a numeral phrase (NumP) which refer to the salient semantic properties of the counted noun (Gil 2013). Numeral classifier systems are the most recognised type of classifier system in the world’s languages. Numeral classifiers are structured by semantic principles, by which certain subclasses of nouns are assigned to the same classifiers. These semantic principles bear certain similarities across the world’s languages, which are called universals. Besides a universals basis, Lyons (1977a: 248) believes that the semantics of numeral classifier are also determined partly by cultural basis. This thesis is a comparative analysis of numeral classifiers in Palikúr (an Arawakan language of Brazil and French Guiana) and Rongga (an Austronesian language). The thesis has two goals. First, I am interested to explore to what extent the semantic universals regarding numeral classifiers hold for Palikúr and Rongga. In particular for Palikúr, since South American languages were mostly left out in the search for universal patterns in numeral classifier languages. The universal semantic frameworks regarding numeral classifiers that I discuss are Allan (1977), Lyons (1977) and Croft (1994). My second goal for this thesis is to explore the semantic variability between the Palikúr and Rongga numeral classifier systems. This question is of interest because it can help illuminate what kinds of semantic differences can be found within the semantic universal parameters identified by semanticists. The semantic variability of numeral classifier systems that I discuss is based on the study of Grinevald (2015). Together, these two questions shed light onto the interplay between universality and variability in the semantics of numeral classifier systems.Show less