Keynotes

Ricardo Bermúdez-Otero, University of Manchester

Title and abstract TBA

Richard A. Blythe, University of Edinburgh

Competition between linguistic variants at the individual and population scale

A number of documented language changes (e.g., names for an object, or alternative grammatical structures) can be couched as a competition between an existing ‘conventional’ variant and one or more incoming ‘innovations’. Mathematical models allow one to determine how processes acting at the individual speaker level (such as matching the frequencies of existing variants, cognitive biases that favour specific innovations or sociolinguistic prestige effects) manifest themselves at the population scale, with social network structures mediating between them. These models have counterparts in population genetics, and one can recognise analogues of mutation, selection and drift in language change. In this talk, I will show that disentangling these fundamental contributions to change is difficult, in part because the simplest neutral theory (in which all changes are random) exhibits rich behaviour. Moreover, it is hard to find combinations that are consistent with empirical data over many languages and long time periods. I will also attempt to offer possible resolutions of these inconsistencies.

Janet B. Pierrehumbert, University of Oxford

Word frequency effects in regular sound change

The classic theory of sound change distinguishes regular sound changes from analogical changes. Analogical changes are categorical changes in the expressions of morphosyntactic categories, typically bringing rare words into conformity with more common words. Regular sound changes instead gradually transform the phonetic realisations of phonemes. The role of word frequency in regular sound changes is actively disputed, with different studies finding frequent words in the lead, rare words in the lead, or no effect.

This talk will review the empirical evidence for word specific effects in regular sound change, and also present an exemplar model that can generate the variety of outcomes reported. The model builds on experimental results about the effects of word frequency and typicality in perceiving speech and laying down memory traces. It goes beyond first-generation exemplar models of sound change by adopting a hybrid approach, in which lexical, phonological, and phonetic levels all play a role.