April McMahon

From speaker to change, from visualisation to causation: quantitative methods and big questions.

Abstract. The sessions in this workshop, which together are intended to take us closer to ‘Building integrated models of linguistic change’, range over change in different levels of the grammar; different causes for change, notably external and internal; and change over different timescales, whether evolutionary, historical, or within the lifespan of the individual speaker. In this talk, I shall focus on how quantitative and computational models can help us integrate divergent viewpoints in these three domains, as well as pinpointing some places where they cannot yet help, and identifying some improvements which might lead us to further advances.

I shall argue that quantitative methods of comparison must crucially allow us to incorporate the results of both internally and externally caused changes. If we follow a strong tradition in historical linguistics and attempt to filter out borrowings and other effects of contact, we are creating an artificial situation and reifying the problematic idealisations of the family tree model. Allowing contact as well as internally motivated changes to be modelled, however, requires us to explore alternative visualisations like networks; but though these do not require any data to be prioritised or excluded, they can be difficult to read and interpret. Moreover, if we are to consider both synchronic similarity or difference and its historical causes, we need to take into account data from as many different areas of the grammar as possible, precisely because in some historical situations, borrowing may affect primarily the lexicon, or in cases of areal linguistics, mainly other domains. We therefore require methods for comparison and associated visualisations which allow us to deal with the consequences of change in different levels of the grammar. Finally, we need models and visualisations which will allow us to work with variation in an individual speaker’s usage, through change in progress across varieties of a language, to its historical consequences in terms of divergent branching at the language level. If change within the lifespan seeds change in progress, and if the increased contact among speakers of different varieties might accelerate change or introduce new targets for changes, then we can no longer validly distinguish methods for modelling and visualising ‘historical’ change from those used in sociolinguistic or dialectological investigation.

These questions demand new methods of comparison; new ways of visualising the outcomes of change; and new ways of interpreting these results. However, they crucially also require greater collaboration between linguists, and between linguists and colleagues from other disciplines. I shall illustrate some possibilities using the example of a method for phonetic comparison which can be applied to the speech of individuals; to the spread of variation across present-day English accents; and to the divergence of modern Englishes from a range of historical varieties. I shall also note a number of cases where we need more data, different methodological approaches, and clearer visualisations to advance work in this rapidly developing area of research.