In various languages, the notional complement to certain attitude verbs is expressed using characteristic morphology of relative clauses -- e.g. Adyghe (Caponigro and Polinsky 2011), Bulgarian (Krapova 2010), Gungbe (Aboh 2005), as well as Nez Perce. The distribution of these "relative embeddings" has a clear affinity for factives. In this talk I investigate the syntax and semantics of relative embeddings with an eye to understanding the connection between relative morphology and factive verbs in particular. I focus on a case study of Nez Perce, for which I argue that relative embeddings do not feature a hidden NP or DP superstructure. Rather, the relative embedding is distinct from its non-relative counterpart in contributing a simple, non-centered proposition -- exactly the type of semantic object most suitable for the calculation of a factive inference. I conclude with a look at variation among relative embeddings across languages, as well as variation in the expression of factives, in light of these arguments.