The overarching goal of the present dissertation was to examine the mechanisms that underlie two word learning paradigms: paired-associate (PAL) and cross-situational word learning (CSWL), when studied via a comparable design. Each paradigm has been developed under different theoretical umbrellas, and thus has been studied separately, limiting the ability to gain a comprehensive understanding of the mechanisms that might be common across them. Four factors hypothesized to affect novel word learning have been investigated in PAL: novel word familiarity, phonological working memory, bilingualism, and delayed testing. In contrast, very few studies have manipulated word familiarity in CSWL, and those that did have used different methods than in PAL. Similarly, the role of phonological working memory in CSWL has not yet been directly investigated. In bilingual studies of PAL and CSWL, a bilingual advantage in learning has been found in PAL, but less reliably so in CSWL. In PAL, this effect has been attributed to better phonological working memory, although findings are mixed. No direct study of the role of phonological working memory has been conducted in bilingual CSWL. Finally, few studies exist in PAL and CSWL examining long-term retention of novel words. We examined the role of word learning paradigm, word familiarity and phonological working memory in word learning in monolinguals in Experiment 1, and across monolinguals and bilinguals in Experiment 2. Experiment 3 examined how characteristics of PAL and CSWL might influence retention of novel words at immediate and delayed testing, and whether phonological working memory would support word learning in the two paradigms differently across different time points. English-speaking monolinguals (Experiments 1, 2, and 3) and native English (L1) - Spanish (second language, L2) adult bilinguals (Experiment 2) were recruited and randomly assigned to either PAL or CSWL (Experiments 1, 2, and 3), and immediate or delayed testing condition (Experiment 3). Additionally, we included two measures of phonological working memory: a backward digit-span and a nonword repetition task. Findings in Experiment 1 suggest that PAL is easier than CSWL, likely due a lower level of ambiguity at learning, and that phonological working memory and word familiarity facilitate word learning across both paradigms. In Experiment 2, results suggest a weak trend for bilinguals to perform better than monolinguals in PAL, with both groups showing similar sensitivity to familiarity and reliance on phonological working memory. The familiarity effect was stronger in CSWL than in PAL, and there was a trend for phonological working memory to be more strongly involved in PAL compared to CSWL. In Experiment 3, performance was higher in PAL than in CSWL, at both immediate and delayed testing, and phonological working memory supported learning to a similar extent across paradigms and testing points. Taken together, these findings suggest that although PAL and CSWL have been studied under drastically different theoretical frameworks, the two paradigms are remarkably similar in their reliance on phonological working memory, and in their sensitivity to novel word familiarity. Additionally, this study tested both monolingual and bilingual adults in PAL and CSWL for the first time and suggests minimal effects of bilingualism on both types of learning, and instead highlights robust consistency in word learning patterns and the mechanisms that enable them across groups. Finally, retention across both paradigms was also tested for the first time and indicates similar reliance on phonological working memory and comparable performance within paradigms over time. Together, the results of this dissertation suggest the need to consider word learning via different paradigms - such an approach can bridge disparate theories of word learning and can ultimately lead to the development of a comprehensive theoretical framework. Practically, these findings show that, whether in the classroom or in an immersion setting, word learning is robust to individual differences and linguistic and timing factors.