Different names are used when people refer to the same entity (e.g., “bread” vs “bagel”).
We built on the ManyNames v2 dataset, a rich resource of real-world images annotated with
multiple English names, to create a Mandarin Chinese version: ManyNames ZH. This new
extension provides an excellent platform for analyzing variations in object naming behavior
within Mandarin-speaking communities. We investigated the relationship between people’s
naming variation and levels of familiarity, and whether ...
Different names are used when people refer to the same entity (e.g., “bread” vs “bagel”).
We built on the ManyNames v2 dataset, a rich resource of real-world images annotated with
multiple English names, to create a Mandarin Chinese version: ManyNames ZH. This new
extension provides an excellent platform for analyzing variations in object naming behavior
within Mandarin-speaking communities. We investigated the relationship between people’s
naming variation and levels of familiarity, and whether this relationship varies in different
ontological domains. Our findings indicate that Familiarity influences naming variation by
two competing forces. Increasing familiarity expands vocabulary and leads to higher
variation, while greater familiarity promotes convergence on conventional names, reducing
variation. The dominant force is convergence, observed across most domains. However,
some evidence of vocabulary expansion is seen in the comparison between PEOPLE and
ANIMALS_PLANTS domains and in the low familiarity range within the CLOTHING
domain.
+
