We explain the typology of sound systems by modelling phonology as well as phonetics bidirectionally (i.e. we model the speaker as well as the listener), and by modelling the acquisition and cross-generational evolution of all this. We either model this in a symbolic framework based on strict constraint ranking (Optimality Theory), or in a distributed framework based on artificial neural networks. If we employ a symbolic framework, we employ at least five representations (one ‘semantic’, two phonological, two phonetic) and four constraint families that connect these representations to each other. We model the processes of comprehension and production and their acquisition and evolution explicitly with computer simulations, and we test aspects of this model by performing laboratory experiments with adults and infants.
The group was joined in March by Camilla Horslund and in September by Mayuki Matsui.
Jeroen Breteler defended his PhD thesis on 30 May 2018. He gives Optimality-Theoretic accounts of the typology of tone spread and tone shift in terms of ternary feet, including computer simulations of its learnability. A major result is that although many unattested patterns are representable in the typology generated from free reranking of constraints, these patterns tend to be less learnable than the patterns actually attested in tone languages.