Valency-changing operations like passivization, causativization, and applicatives alter predicate argument structure. Passivization demotes agents and promotes objects; causativization adds external causers; applicatives add beneficiaries. These operations are often expressed morphologically and illuminate the lexical-syntactic structure of predicates, showing that argument structure is not fixed but systematically transformable through dedicated operations.
From argument structure and thematic roles, you know that verbs come with a fixed set of participants they require — the verb *give* demands a Giver, a Recipient, and a Given-thing; *break* needs an Agent (optionally) and a Theme. This argument structure is stored in the lexicon as part of the verb's meaning. Valency-changing operations are the grammatical machinery that alters this stored structure in principled ways, either adding, removing, or reorganizing participants without changing the core event being described.
Passivization is the most familiar example and the clearest illustration of the logic. Take *The thief broke the window*: Agent (*thief*) and Theme (*window*), with Agent as grammatical subject. Passivize it: *The window was broken (by the thief)*. The event is the same; the arguments are the same; but the Theme has been promoted to subject position and the Agent has been either demoted to an optional *by*-phrase or suppressed entirely. The valency — the number of obligatory arguments — effectively decreases: in the passive, the Agent can be omitted, whereas in the active it typically cannot be. Passive morphology is the surface signal that this structural reorganization has occurred. Languages vary enormously in how they encode this: English uses an auxiliary (*was*) plus a participial form; Latin uses morphological endings on the verb; many languages use a reflexive clitic.
Causativization works in the opposite direction — it adds a participant rather than demoting one. The verb *melt* is intransitive in *The ice melted* (just one argument, the Theme). Causativize it: *She melted the ice*. Now there's an Agent (*she*) who caused the melting, plus the original Theme. The valency increases by one. Many languages encode this morphologically: in Japanese, adding the suffix *-sase* to a verb causativizes it, and in Turkish, *-dır/-tır* serves a similar function. What makes this theoretically important is that causativization doesn't simply add any participant — it adds specifically a causer who stands outside the original event and brings it about. The morphology tracks a semantic relationship, not just an extra noun.
Applicatives add yet another type of participant: a beneficiary, instrument, location, or goal that would otherwise be expressed as an oblique or adjunct. In English, *She baked him a cake* is a dative alternation — *him* is semantically a Recipient but syntactically an object. In languages with applicative morphology (many Bantu languages, Quechua, some Mesoamerican languages), a dedicated affix on the verb promotes an oblique participant into the core argument structure, giving it direct object properties like agreement and case. This matters because it shows that what counts as a "core argument" is not just a semantic question — it's a language-specific grammatical choice that morphology can modulate.
The unifying insight across all three operations is that argument structure is two-layered: there is the underlying semantic structure (who did what to whom), and there is the grammatically projected structure (which participants are subjects, objects, or obliques). Valency-changing operations manipulate the mapping between these layers. This view predicts that you can have a single semantic event — breaking, melting, giving — realized with different grammatical argument configurations depending on which morphological operations have applied. It also predicts that these operations should interact systematically: you can passivize a causative (*The ice was melted by her*) because the same layered architecture applies at each step. Recognizing these operations is essential for analyzing any language with rich verbal morphology.