Our speech is affected by recent naming experience (i.e., what we have said before). For example, after talking about your cat with friends, you may mistakenly say “cat” when you intended to say “dog”. Although the influence of naming experience on speech production has been known for a long time, it is unclear how naming experience shapes the language system and thus affects future speech. Specifically, whereas some studies find that naming semantically related pictures speeds up subsequent naming, recent studies demonstrate that previously naming semantically related pictures slows down future naming. Because these studies use various paradigms with different materials and experimental parameters to investigate how past naming influences future naming, it is difficult to explain how the same naming experience results in opposite effects (i.e., facilitation and interference). The goal of this dissertation is to bridge the gap between these contrasting results and determine how the language system is changed by past naming experience. To accomplish this goal, I tested both semantic facilitation and interference in naming within the same paradigm. In Experiment 1, by manipulating the interval between two naming events, I established the opposite effects caused by semantically related naming experience, suggesting that naming experience has two temporally distinct effects on subsequent speech production, short-lived facilitation and long-lasting interference. In Experiments 2 and 3, following previous methodology while simultaneously addressing its limitations, I investigated the cognitive locus of semantic facilitation and interference respectively. Results suggest a role of semantic processing in both facilitation and interference. Lastly, to provide further and convergent evidence about the locus (semantic vs. lexical) of semantic facilitation and interference in naming, I used fMRI in Experiment 4 to explore the neural loci of these two effects in the same picture naming experiment with one group of participants. The results suggest that facilitation has a semantic locus, while interference has loci at both semantic and lexical levels of processing. Together, these experiments systematically investigated the mechanisms and loci of facilitation and interference caused by past naming experience from both behavioral and neural perspectives. This work sheds light on the dynamic nature of the language system providing insight into how our past experiences shape our current cognitive processes/abilities.