In the phonological integration of loanwords, the original structures of the donor language can either be adopted as innovations or adapted to the recipient language. This dissertation investigates how structural (i.e. phonetic, phonological, morpho-phonological) and non-structural (i.e. sociolinguistic and psycholinguistic) factors interact in determining which of these two integration strategies is preferred. Factors that affect the accuracy of the structure’s perception and production in the donor language as a result of its acquisition as a second language are given special consideration. The three studies in the dissertation examine how the same phonological structure from different donor languages is integrated into the same recipient language Turkish by two different types of initial borrowers: elite bilinguals in Turkey and heritage bilinguals in Sweden. The three investigated structures are word-final [l] after back vowels, long segments in word-final closed syllables, and word-initial onset clusters. The main hypothesis is that adoption will be more prevalent in heritage bilinguals than in elite bilinguals. Four necessary conditions for adoption are identified in the analysis. Firstly, the donor-language structure must have high perceptual salience. Secondly, the borrowers must have acquired the linguistic competence to produce a structure accurately. Thirdly, the borrowers must have sufficient sociolinguistic incentive to adopt a structure as an innovation. Fourthly, prosodic structures require higher incentive to be adopted than segments and clusters of segments. The main hypothesis is partially confirmed. The counterexamples involve either cases where the salience of the structure was high in the elite bilinguals’ borrowing but low in the heritage bilinguals’ borrowing, or cases where the structure’s degree of acquisition difficulty was low. Therefore, it is concluded that structural factors have the final say in the choice of integration strategy. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 3: Submitted. </p>