Kinyarwanda noun classes
Starting with behavioral data, we explored whether Kinyarwanda speakers were sensitive both to morphological and semantic composition of nouns when asked to find the 'odd one out' in a group of three. Preliminary results suggest that speakers are sensitive to both dimensions, and that individuals differences and additional languages spoken shape how speakers perform in this task. (See our recent talk at Bantu8!)
Next we aim to gather property norms from Kinyarwanda speakers, which will provide data for a computational analysis of semantic cohesion within noun classes. Using representational dissimilarity analysis (RDA), we intend to quantify the distances between nouns in any given noun class, based on shared semantic properties. Stay tuned!