For years, second language teachers and practitioners have known that commercially produced language teaching textbooks provide inauthentic language samples, which sound unnatural or stilted, and do not represent spontaneous pragmatic and real social language used in natural conversation (Ishihara and Cohen 2021, House 2014, Kasper 2014, among others).  In this context, the second language learner is rarely exposed to many of the Spanish varieties that they are bound to encounter, either at home, in their communities, or in potential study abroad programs. To address such a void, we at Princeton University have built a corpus of spoken Spanish featuring speakers of different varieties of “Spanishes”, in both interpersonal (real spoken interactions) and presentational modes (interviews).  In this presentation, we will describe the process of building our corpora (Voces de Princeton and Intercultural Spoken Interactions) and how we use our data sets for pedagogical purposes at different levels of Spanish proficiency. We will delineate the steps to corpus-based materials development, addressing different students’ needs, and targeting different learning goals in selected excerpts such as the use of pragmatic features, language form-function samples, or dialectical variations (mainly, lexical and phonological differences). We will showcase some focused-noticing tasks that guide our students in learning pragmatically and sociolinguistically appropriate uses of the target language so that they know how to use them in authentic scenarios.