Against past analyses, we propose that natural language causatives do not universally encode a single, unanalyzable bringing about meaning like Dowty’s (1979) CAUSE, but instead draw on an inventory of contrasting causal dependency relations. To illustrate this claim, we focus on the English causative verbs make and cause. We point out a number of differences in their inferential profiles, and argue that these follow from the fact that cause asserts a relation of causal necessity between a cause and its stated effect, while make asserts causal sufficiency. We distinguish these notions from their alethic counterparts: while causal necessity is similar to the notion of counterfactual necessity (Lewis 1973), causal sufficiency has not figured in previous analyses of causal language. We show that analyzing make as a sufficiency causative not only accounts for the similarities and differences between its distribution and that of cause, but also enables us to explain previously puzzling inferences associated with the use of make as opposed to other periphrastic causatives.

Past-tense ability ascriptions (e.g., "was able") alternate between pure, potentially unrealized ability and an interpretation which actualizes the ability. The alternation extends to abilitative uses of possibility modals, with actualized readings strengthening to actuality entailments under perfective aspect in aspect-marking languages (Bhatt 1999). These interpretations resist explanation on accounts which seek to derive them in the composition of modality and aspect. I build on causal analyses of implicative inferences (from lexical implicatives like "manage" as well as 'variably-implicative' "enough" comparatives; Baglini & Francez 2016, Nadathur 2016, 2017) to propose a new approach to actuality inferences, grounded in a causal semantics for ability predicates. This account derives both pure ability and actuality readings, and explains parallels between implicative "manage" and actualized ability on the basis of shared (presuppositional) causal structure. "Manage" and ability differ in asserted content, but the difference is neutralized – producing actuality entailments – under a perfective operator which selects for eventive predicates, and combines with stative ability attributions only via aspectual coercion.

Karttunen's (1971) implicative verbs are notable for generating inferences over their complements. English manage to X, for instance, entails the truth of X: the entailment reverses with matrix negation and seems tied to the elusive presuppositional contribution of the implicative predicate (Coleman 1975). Building on Baglini & Francez (2015), and drawing on implicative data from Finnish, I propose an account of the implicative class which links the lexical presuppositional content of an implicative verb to inferences over the truth-value of its complement via a model of causal relationships between contextually-salient variables (Schulz 2011). Unlike previous proposals, this account extends naturally to the commonalities between manage and weaker one-way implicatives like Finnish jaksaa(=have.strength), which only entail under one matrix polarity.

No abstract

No abstract