Does economy (or efficiency) explain grammatical change?

In this big-picture talk, I address the question whether cost-benefit tradeoffs (often known under the label of communicative efficiency, Levshina 2023) have a central role in explaining language change. Since Gabelentz (1891), linguists have often made claims along these lines (e.g. Langacker 1977; Keller 1994), but these ideas have generally remained speculative and have been supported by individual examples and intuitive judgements, not by systematic cross-linguistic evidence. The purpose of this talk is to throw serious doubt on this time-honored idea.

       The term “economy” has also been used in the sense of “system economy”, a principle that favours minimization of elements of the system rather than a tradeoff between costs and benefits in actual communication (e.g. van Gelderen 2004). I will briefly comment on some of these proposals, pointing out that there is no clear relationship between system economy and communicative efficiency. Quite generally, the idea that system considerations drive or explain language change suffers from the problem that change in language use must always precede system change (as emphasized e.g. by Lightfoot 1999).

       That communicative efficiency shapes language structures to a substantial extent is now widely recognized (e.g. Gibson et al. 2019), and extensive cross-linguistic evidence has been assembled for various domains of grammar. There is also general agreement that language change must play an important role in bringing about efficiently designed systems, as in biological change (e.g. Croft 2000). But is it the nature of the diachronic processes and pathways of change that is responsible for the resulting efficient language structures? This has been argued in recent years (e.g. Bybee 2006; Cristofaro 2019), and if true, this would be in line with the old idea that efficiency tradeoffs instigate and drive the change processes.

       Here I argue, by contrast, that the causal relationship is the reverse: Language users unconsciously prefer efficient variants in language use, which results in overall efficient systems. The changes that lead to the resulting systems have very similar results, but their starting points and trajectories are very diverse. I will give examples from a range of languages (mostly Indo-European) in the following domains of grammar: (i) differential object marking, (ii) adnominal possessive marking (cf. Stolz et al. 2008); (iii) future tense marking; (iv) plural marking; causative and anticausative marking (cf. Haspelmath 2016). Establishing causality in language change is generally very difficult (if possible at all), so the argument is primarily built on plausibility considerations.