Understanding the relationship between syntactic structures and linear strings is a challenge for modern syntactic theories. In recent years an interesting new hypothesis has begun to emerge. Bennett et al (2016) discusses an anomalous displacement in Irish in which prosodically-light pronouns are displaced to the right of their expected position, with no change in meaning. This appears to be evidence that the linearization procedure does not operate purely on syntactic structure, but rather needs to know the phonological form of individual items in order to order them. I term this phenomenon prosodic displacement. In this dissertation, I describe a new case of prosodic displacement in Khoekhoegowab, a Khoisan language of Namibia, which causes light tense-marking auxiliaries to displace leftward. Based on this data and the other known cases of prosodic displacement, I propose a theory of Optimal Linearization, which models linearization using violable constraints at the syntax-prosody interface. These constraints alone give us new insight into the linearization process: The fact that specifiers are always on the left is modeled as an emergence of the unmarked preference for head-finality, while the Final-Over-Final Constraint is captured using a domain-specific head-finality constraint. The interaction of these linearization constraints with other specifically-prosodic constraints results in prosodic displacement whenever the 'expected' order would yield a marked prosody.