This thesis aims to explore how psycholinguistic factors can play a pivotal role in the way that the human lexicon is structured. Recent approaches to studying language have been heavily influenced by the principle that languages are shaped to suit the general learning and processing biases of the learner (Christiansen & Chater, 2008). This view has gained considerable support from theoretical and empirical accounts from researchers in the fields of language acquisition and evolution. However, there has been little evidence that demonstrates how such learning biases may operate differently at various stages of an individual’s vocabulary development, or indeed how they may operate differently as the vocabulary subsequently evolves. Through a series of experimental studies, this thesis will examine how well documented psycholinguistic properties of words can shape the lexicon throughout the course of a language learner’s life, as well as the life of the language itself.Chapters 2 and 3 investigate how both arbitrariness and non-arbitrariness (in the form of sound-symbolism) make significant contributions to the way that languages are learnt. Central to this point is that important considerations should be made as to when and how these two properties are considered beneficial. Principally, these chapters focus on how the size of the vocabulary directly influences whether arbitrariness or non-arbitrariness is advantageous for the learner. We show that sound-symbolism facilitates learning individual form-meaning mappings when the vocabulary size is small, whereas as the vocabulary grows this advantage is only observed for distinguishing between categories. Which may explain why words acquired earlier in life, when the vocabulary size is small, are reported to be more sound-symbolic (Monaghan et al, 2014).Chapter 4 goes on to examine how variation in psycholinguistic properties of word can be used to directly predict variation in the fidelity of the word’s learning and production. By manipulating the frequency, length and age of acquisition of words learnt within an artificial language learning paradigm, this chapter aims to show how certain words are less vulnerable to change based on the way these properties offer significant processing advantages, whilst also considering the nature of the way errors are produced during recall by distinguishing between lexical adjustment and replacement. Importantly, these properties are investigated independent of one another, allowing for a significant contribution to be made to the way language change is studied.Following on from this, Chapter 5 then considers how these psycholinguistic properties may come to change the lexicon over the course of evolution. By using an iterated learning paradigm, the aim here is to observe how the changes reported in Chapter 4 may explain previously reported differences in the stability, and indeed instability, of lexical forms over a much longer timescale. This approach examines such differences through the cultural transmission of the languages across several generations of learners.