One of the most discussed questions in the literature on sound change is whether functional factors play a role in shaping the lexicon over time, for instance by blocking the occurrence of a sound change (Gilliéron 1918), or whether sound change is entirely mechanical and phonetically determined (Labov 1987). Interestingly, the functional factors that have been proposed in the literature, like the principle of least effort (Zipf 1939) and the minimization of entropy loss (Hockett 1967), have not been related to the literature on language acquisition, with few exceptions.In this work, I address this question by formalizing a model of sound change over time, in order to compare its predictions with the findings of historical and contemporary investigations on sound change (Chapter 2), by revisiting the hypothesis that sound change is blocked when it leads to homonymy (Chapter 3), and by identifying the factors that predict phonological development in children (Chapter 4). While in the first part of the dissertation I argue that lexical change over time can be modeled without reference to functional considerations (contrary to the models in Martin 2007, Graff 2012 and Dautriche et al. 2017), in the second part I focus on the proposal that sound change seems not to occur when it would lead to homonymy (the 'Functional Load Hypothesis' in Martinet 1955), and I argue that while this might be true for sound changes that result in the loss of a sound contrast in the speaker's grammar, it is not true for other common types of sound change. I motivate this hypothesis by showing that lexical contrast is a factor that influences early phonological development.