1. Introduction
The (anti)causative alternation is a transitivity alternation concerned with how languages express externally caused (CAUSAL) vs. spontaneously occurring (NONCAUSAL) events. Typologists have shown that languages resort to three main coding patterns to express this alternation (Tubino-Blanco 2020):
(i) CAUSATIVE: the causal member is overtly marked, Turkish öl- ‘die’ vs. öl-dür- ‘kill’
(ii) ANTICAUSATIVE: the noncausal member is marked, Italian scioglier-si ‘melt (intr.)’ vs. sciogliere ‘melt (tr.)’
(iii) EQUIPOLLENT: both members are equally marked, Yaqui (Uto-Aztecan) bee-te ‘burn (intr.)’ vs. bee-ta ‘burn (tr.)’
Despite the structural parallelism, it has been pointed out that causative and anticausative markers (AMs) show a cross-linguistically unbalanced distribution. Not only are AMs typologically less frequent than causatives, but it is also rare for AMs to outnumber causatives in the lexicon of individual languages (Nichols et al. 2004). In fact, the very existence of the anticausative pattern, in which the semantically simpler member of the alternation is morphologically more complex, constitutes a puzzle, as it violates the iconicity principle (Haspelmath 2016: 593).
Within individual languages, the distribution of AMs is not random, as some verbs trigger AMs more frequently than others. To explain these distributions, scholars have resorted to either verb semantics or frequency effects. Semantics-based accounts appeal to notions such as spontaneity (Haspelmath 1987) and claim that verbs lexicalizing events less likely to occur spontaneously are more likely to trigger AMs, because higher cognitive markedness entails higher structural markedness (Haspelmath 1993: 106). Frequency-based approaches explain marking asymmetries as mirroring frequency asymmetries, based on the assumption that higher usage frequency items are more predictable and thus favor shorter coding (Haspelmath 2021). This means that verbs that more routinely occur in noncausal contexts are less likely to receive AMs (Haspelmath et al. 2014).
In particular, Haspelmath (2016) proposes that verb meanings can be ranked according to the spontaneity scale in (1), which predicts that if a language has AMs for a specific class of verbs, it also has them for all verbs lower on the scale.
(1) TRANSITIVE ‘cut’ > UNERGATIVE ‘talk’ > AUTOMATIC UNACCUSATIVE ‘melt’ > COSTLY UNACCUSATIVE ‘break’ > AGENTFUL ‘be cut’
Overall, despite the existing research, the fundamental question as to why anticausatives exist remains open. With this project, we aim at building an international network of scholars, in order to develop a new diachronically-oriented approach to the study of AMs. The project is articulated into two main work packages (WP).
2. WP1: the diachronic typology of AMs
To understand why anticausativization exists, one must explore how AMs arise in the first place. This WP focuses on the diachrony of AMs cross-linguistically, by tackling two main questions.
The first research question concerns the cross-linguistic sources of AMs. Typological works mention only two such sources, reflexives and passives, but there is evidence for alternative developments (Bahrt 2021). We therefore set out to explore the range of non-reflexive sources of AMs. Second, by adopting a source-oriented perspective (Cristofaro 2019), we investigate to what extent the recurrent synchronic distribution of AMs can be explained by considering the diachronic sources and processes whereby they emerge. The hypothesis is that individual sources of AMs may give rise to either anticausativization or equipollent patterns, and this may explain why the anticausativization pattern is overall rare. In addition, we also explore whether cross-linguistic diachronic data supports the existence of the spontaneity scale in (1). Available diachronic evidence does not fully match our expectations: AMs rarely originate