Languages are vastly different. One example of striking variation is the way that languages categorise nouns. Ukrainian assigns its nouns to one of three genders, while Fula has twenty different genders, and Finnish operates without gender. To a native Finnish (or English) speaker, gender is baffling. Yet over 40% of the world’s languages make use of such systems of nominal categorisation (Corbett 2005). In the Oceanic languages spoken in the South Pacific, nominal categorisation manifests itself in a very different way — through possessive classifiers. When a person uses a possessive pronoun, i.e., ‘my’ or ‘yours’, they must evaluate the possessed item in terms of how they interact with it. A special classifier word is used — so speakers have to say: ‘my FOOD yam’ or ‘your DRINK water’, where FOOD and DRINK are the possessive classifiers. Typically, these classifiers are not like a fixed grammatical gender system in Ukrainian where ‘water’ is always feminine. If the possessor drinks their water, then the DRINK classifier is used; whereas if they wash with it, then a different classifier, such as the GENERAL classifier, is used.

These possessive classifier systems are relatively rare. They are intriguing, and many of the languages that have them are endangered, with speaker populations between 400 and 6000. To learn more about these systems we are comparing possessive classifiers in six Oceanic languages: Merei, Lewo, Vatlongos, North Ambrym (Vanuatu), Nêlêmwa and Iaai (New Caledonia). Each has a different set of classifiers, from a simple two-way distinction (Merei) up to a more complex inventory of 23 (Iaai). We have designed a novel suite of psycholinguistic experiments to compare these different systems, making significant discoveries about their differences and about how free categorisation can turn into rigid classification. This sheds light on the origin of gender systems, and the very nature of gender in terms of how humans categorise entities within the world. This is helping us to reveal how the mind codifies human experience, and how language contributes to this fundamental process.

We are tracking language use on the ground, in the moment, and are uncovering both predicted and surprising findings (see e.g., Grandison et al., 2021). Strikingly, the results from our free-listing, card sorting, video vignette, possessive labelling, eye-tracking and storyboard experiments show that there is considerable inter-speaker variation in the use of classifiers. Whilst there is some evidence for an increase in the use of the GENERAL classifier over more semantically precise classifiers, we are also revealing the existence of several new classifiers, not previously documented. For example, in Iaai, a language formally recognised as having 23 classifiers, we see use of at least an additional 11. This demonstrates again the productive and dynamic nature of language. What is not yet clear is how consistent and widespread the use of these new classifiers is and the impact of their usage on cognition. We will discuss this intriguing emergence and how our work is addressing these questions.