This dissertation investigates the IP-related categories and how the verbal domain of Turkish is organized. Within the theory of Universal Grammar, there are three major approaches to the IP domain of languages. The initial distinction is between the Syncretic IP model and the Rich IP model. The former refers to the conception that human language only makes available the heads and phrases required in a specific derivation, and that languages display parametric variation while the latter argues that the human mind comes with a highly articulated and rigid schematic hierarchy where all features of all functional categories are available in every derivation without parametric variation, yet most of them are silent. Additionally, there is an intermediate hypothesis, the Split IP model, where only major categories such as tense, mood and aspect, have dedicated head positions, and the morphological form inserted to each head position specifies its value. The dissertation aims to find out which one of these models is supported by the data in Turkish. I argue that split or syncretic character of the IP in Turkish should be sensitive syntactic operations that can target the functional heads individually. With this in mind, I suggest that a non-finite adjunct clause in Turkish is exceptional in that it lacks any kind of content when it stands alone, and therefore cannot be uttered in isolation. Yet when adjoined to a matrix clause, it is interpreted as having the values of the functional heads in the matrix clause via the mechanism ‘copy’. The data illustrates that although ‘copy’ can target some heads individually, there are two sets of heads that are always copied as a whole. Assuming that ‘copy’ can only single out independent heads, I conclude that Turkish has two syncretic phrases where two morphemes co-head the phrase. Specifically, ability modal and negation form the deontic modality phrase (DmodP) while tense co-heads another phrase with an aspect or modal marker (TAMP).