Abstract

In this paper I address the diachronic evolution of pragmatic markers, with a focus on what I call linear vs non-linear forms of pragmaticalization. The rise of pragmatic markers has been an increasingly popular research topic for more than three decades. The most frequently attested pathway involves linguistic items or constructions that originally have truth-conditional meaning and belong to « core » grammar, but which more or less gradually evolve non-truth-conditional, more (inter)subjective, uses that lie outside « core » grammar. This form of evolution is widely assumed to be regular, unidirectional, and thus fundamentally linear in nature. Saliently, the study of such cases been used to argue for a redefinition of the notion of grammaticalization. The literature has, however, reported examples of markers that appear to have taken more complex, non-linear, paths at the semantic-pragmatic and/or the syntactic level. These types of cases have been/are the focus of two international research networks of which I was/am the PI : the British Academy-funded « Cyclicity in Semantic-Pragmatic Change : from Latin to Romance » (2017-18) and the AHRC -funded « The Role of Pragmatics in Cyclic Language Change » (2021-23). In my talk, I argue that because pragmatic markers may evolve along a variety of non-linear – including but not limited to cyclical – paths, it is not helpful to subsume the rise of pragmatic markers under the concept of grammaticalization. Instead, in order to arrive at a descriptively adequate account, it is more useful to draw on a distinction between grammaticalization, pragmaticalization and lexicalization. I identify four non-linear forms of pragmaticalization, based on attested patterns of evolution prominently involving interaction between, on the one hand, two levels of meaning (the Content Level and the Context Level), and on the other hand, two levels of grammar (Micro-Syntax and Macro-Syntax). My examples will be adduced from several Romance languages.