This study concerns word-formation in Turkish within the scope of theoretical morphology. I propose a paradigmatic analysis of Turkish Noun-Noun compounding and subsequent derivation within a process-based approach to morphology. I propose that Noun-Noun compounds (NNCs) such as kar ışıltı-sı (snow brightness-sI) ‘brightness of snow’ are produced by the morphological component, which is separate from both lexicon and syntax, and describe the nature of morphological operations in Turkish. This involves accounting for the semantic combination and the structure of NNCs, including the presence of the compound marker underlined above. I propose that the compound marker appears in order to create a lexeme from an N-N stem, which signals that a word-formation process has taken place. This makes -sI a derivational suffix in NNCs. I also propose that the compound marker is in a slot competition with certain derivational suffixes on the basis of non-permitted sequences in derivations involving compounding, e.g. kar tane(*-si)-li(*-si) (snow flake-(*-sI)-lI(*-sI)) ‘with snow-flake’. Namely, the compound marker and certain derivational suffixes are involved in a paradigm structure of word-formation. The proposed paradigm structure highlights the importance of paradigmatic selection in word-formation and accounts for permitted and non-permitted sequences. I also propose that the semantic rule responsible for the associative semantics of NNCs maps onto the compounding of the two nouns rather than -sI suffixation, which follows it. I further propose that the same form rule suffixing -sI also applies in possessive constructions (PCs), e.g. kar-ın ışıltı-sı (snow-GEN brightness-sI) ‘the brightness of (the) snow’. This accounts for why the same formal element appears in two distinct structures and why the sequence *-sI-sI is not allowed. Unlike the compound marker, the function of -sI in PCs is to mark the possessum: -sI in PCs is inflectional. A process-based approach to morphology, which assumes the separation of form and meaning, and treats affixes as rule elements, can easily deal with such cases. Further, I claim that -sI in PCs does not mark (3rd) Person but marks only the value ‘possessed’ of the grammatical category Possessedness in Turkish: it distinguishes possessed vs. non-possessed entities rather than 3rd vs. 1st/2nd Persons.