Abstract:

This paper develops a formal methodology for capturing and representing the semantics of causal expressions in natural languages. Focusing on two causative constructions—covert causatives (change-of-state verbs) and overt causatives (the verb cause)—it provides a proof of concept for analyzing the distinguished meanings of different causative constructions.
We adopt the formal framework of Structural Equation Modeling (SEM) to analyze causality and integrate it into model-theoretic semantics for interpreting causal statements. In our approach, the selection of a cause within a particular construction depends on its inclusion in a sufficient set of conditions that bring about the effect, as well as on specific properties of the cause itself. To formalize this process, we introduce the concept of causative-construction selection (CC-selection), which captures how speakers select a causative construction that aligns with the relational structure between states of affairs. For each relevant condition within the sufficient set, CC-selection determines whether it can be encoded as the cause in a statement articulated
through a specific causative construction, thereby describing a particular state of affairs. We argue that CC-selection plays a central role in shaping the meaning of causal statements.
By leveraging the SEM framework, CC-selection effects can be formally explained through contrasts within the structure of a model. For instance, notions of sufficiency and necessity, which play a crucial role in these selections, are rigorously defined within SEM, allowing for a precise
account of CC-selection effects. This paper further illustrates how CC-selection accounts for contrastive inference patterns across constructions. By focusing on the two causative constructions central to our discussion, it resolves longstanding puzzles associated with change-of-state verbs.
The proposed framework establishes a foundation for the systematic study of causal language, bridging semantics and philosophy while providing tools to investigate the interplay between causative constructions and their associated causal meanings.