A chain shift (CS) is a kind of sound shift in which several individual sound shifts form a “chain”, e.g. /A/ → /B/ → /C/ → [D], where /A/ moves towards /B/, /B/ moves towards /C/, and /C/ moves towards a previously unoccupied position [D]. Noteable examples of CSs are Grimm’s Law (the “Germanic sound shift”, e.g. /bh/ → /b/ → /p/ → [f]), the Great Vowel Shift (e.g. /aː/ → /ɛː/ → /eː/ → /iː/ → [aɪ]) or the on-going Northern Cities Shift (/ɪ/ → /ɛ/ → /ʌ/ → /ɔ/ → /ɑ/ → /æ/ → [ɪa]). Of course, it would be foolish to assume that the partaking changes are not in any way interconnected.
Now, Exemplar Theory (ET) may shed light on the causal links between the individual shifts of a chain shift (or even on sound change in general). ET is a cognitive theory of category formation and judgement. Its main claim is that individual percepts, exemplars, rather than mere abstract representations are stored in long-term memory and that it is clouds of many similar exemplars that form (the illusion of) categories. At any point in time, exemplars may fade from or enter memory, thus making categorical change an inherent aspect of the theory. And by regarding phonemes as clouds of similar phones, it becomes apparent why ET is well-suited to explain phonemic change.
The talk will present and discuss a computational model that aims to simulate the effect of phonetic variation on sound inventories within an ET setting. We will see that, given the right premises, CSs may indeed arise from the dynamics that result from constant exemplar update.
Note: The talk is crafted to be accessible to students of all fields. Only a general clue of what is going on in linguistics at all is necessary.