This dissertation investigates how syntactic, semantic, and pragmatic factors interact to produce the Internally-Headed Relative Clause (IHRC) construction in Korean and Japanese. The IHRC construction differs from the more familiar Externally-Headed Relative Clause (EHRC) construction in several ways. First, unlike an EHRC, an IHRC's content restricts the matrix clause's content, rather than the semantic head's. Second, its interpretation is heavily influenced by the discourse context in ways not seen with the EHRC. Third, unlike the head of an EHRC, the head of an IHRC does not correspond to any overt syntactic phrase, so it needs to be determined by language users, based on the relative clause's content, the matrix predicate's semantics, and the discourse context. The literature offers numerous insightful analyses of the IHRC construction, but it leaves two central questions unanswered: what determines the interpretation of the construction? And, if pragmatic principles have a role to play, how do they interact with the morphosyntax and the semanties? I answer these questions within an event semantics framework. 1 show that the construction's interpretation is determined partly by grammatical factors (e.g., the embedded clause's aspect, the matrix predicate's semantics) and partly by pragmatic factors (e.g., the discourse context and discourse participants' world knowledge). In particular, I isolate two sources of the semantic variability of the construction. First, the matrix clause contains a pronominal definite description, whose denotation contains a free relation variable. This variable's value is determined by the embedded clause's event structure, the matrix predicate's semantics, and the discourse context. Second, the relative operator that occurs in this construction connects the embedded clause's content with the matrix clause's content by establishing either a temporal or a causal relation between them, depending on whether the embedded clause describes a temporary state or a permanent state. This study establishes important connections between the semantics of a definite description and event structure, thereby solving the particularly challenging formallinking problem, one that afflicts existing E-type pronoun analyses. It also provides a constrained but flexible interpretive mechanism for the construction, eliminating the need for many of the extra-grammatical constraints that characterize existing treatments.