Constructional attrition in a radically usage-based model of language (change)

Abstract

Situated in the burgeoning discipline of diachronic construction grammar (cf. Barðdal et al. 2015, and Noël and Colleman 2021), this paper aims to contribute to a radically usage-based (Noël 2016) theoretical account of the phenomenon of ‘constructional attrition’, which is the term coined by Colleman and Noël (2012) to refer to the, either fully accomplished or only incipient, gradual disappearance of a construction from a language. The approach taken is to explore how the phenomenon can be accommodated in Schmid’s (2020) Entrenchment-and-Conventionalization model (or EC-model). This theory of “the dynamics of the linguistic system” is ‘radical’ in its usage-based account in that it separates collective conventionalization from individual entrenchment and models how they interact in usage. Schmid focuses on the accumulative dynamics of the system, however, and gives scant attention to subtractive/ attritional developments. Considering data from two corpus studies on the development of the so-called DEONTIC NCI construction (SUBJ BE obliged/forced/permitted/… to INF) (Noël 2008, and Disney 2016), one on its language-level decline during the Late Modern English period and one on how its speaker-level use evolves over the writing careers of a number of Late Modern English authors, the paper brings to bear the EC-model’s conceptual apparatus to gauge the role of individual speakers in language-level constructional attrition. It is argued that while a sociopragmatic factor can be identified that is likely to have acted as a ‘force’ on usage and conventionalization, contributing to a ‘reversal’ in the conventionalization of the schema, there are good reasons to hypothesize that reversed ‘routinization’/ entrenchment in individual speakers is the leading process in its language-level attrition rather than collective reversed ‘usualization’. These reasons pertain to individual differences of two sorts: in the second of the two studies referred to, a sizeable group of authors maintains a stable level of entrenchment of the schema that remains below the corpus average throughout their writing careers, and another group of authors displays internal attrition of the construction to a level below the corpus average, i.e. they exhibit a decrease in their level of entrenchment of the construction that parallels the external development. The combination of these observations suggests that constructional attrition, at least the kind illustrated by the example of the DEONTIC NCI schema, is in large part the effect both of what Labov (1984) calls ‘generational change’ and of what Sankoff (2005) terms ‘lifespan change’, rather than being a case of what Labov (1984: 84) labels as ‘communal change’, in which “all members of the community alter their frequencies together”. In Schmid’s EC-model, the latter would be language change in which the conventionalization process of (reversed) usualization takes the lead, while both generational change and lifespan change are led by (reversed) routinization. Schmid’s radically usage-based model can therefore equally usefully be applied to account for subtractive developments as for accumulative ones.