Recent psycholinguistic research has explored language switching in bilingual populations to understand how language is stored in the mind. Whereas the bilingual analysis has been useful, multilingual switching may provide more nuanced details about how the multilingual lexicon (vocabulary) is cognitively organized and how languages interact during speech. This dissertation examines the structure of the multilingual lexicon in two ways: (1) reaction time and carrier phrase duration measures are examined to provide evidence of trilingual lexical access when language switching in connected speech; (2) cross-linguistic influence in trilingual production is examined as evidence of the organization of the lexicon. The independent variable is language dominance. This dissertation is the first known study in which the L1, L2, and L3 are purposely varied for participants, which allows for generalizable results regardless of a specific language combination. It is also the first study to use a task that combines sentence reading in conjunction with cued picture naming, where the picture represented the object of the sentence to be produced. There were eleven sets of language triads (e.g. English, Russian, Spanish or English, French, Portuguese); within a triad, eight easily identified words with comparable mean frequencies in each language were chosen as object stimuli. Each stimulus contained a written carrier phrase in one of the participant's three languages. The carrier phrase language was randomly selected, and it was followed by an object represented by a picture, which was to be named in the target language of the session (L1, L2, or L3). Stimuli were presented in conditions; a condition was a set of trials in which the target language of the language switch was always the same. Conditions were counterbalanced, with respect to order of acquisition, so that an equal number (n = 7) of the total participants (N = 42) followed each of the six potential orderings, e.g. L2, L3, L1 or L1, L3, L2, which was repeated for the total of nine conditions. In each condition, thirty-six stimuli were presented. All forty-two participants also completed a language profile questionnaire in order to gauge language dominance. Results from three mixed models (two for each of the reaction time analyses and one for the carrier phrase duration analysis), examining the main effects, support prior studies of lexical access investigating switch costs in which switching into a more dominant language exhibits a slower reaction time. Additionally, the results provide complementary approaches where latencies may be attributed to the predictability of a language switch. The cross-linguistic influence results partially corroborate past research in which more dominant languages were more easily suppressed, but do not strongly support the claim that more dominant languages cause less cross-linguistic influence. Rather data show that the L2 is both the most influential and most affected language, which supports the L2 Status Factor Model which states that the L2 has an intermediary position between one's native language and other languages. Considered together, a more nuanced picture of the multilingual lexicon appears in which forces of inhibition and activation exist in an interplay to moderate lexical selection.