This dissertation is an empirical investigation of mitigation strategies in spoken Spanish between L1 speakers and advanced Swedish L2 learners. The first aim of our study is a pragmalinguistic analysis and description of the mitigating functions of a number of linguistic elements in our corpus. The second aim comprehends a study of three contrasts, namely, the use of mitigation strategies in (a) two speaker groups: L1 vs. L2 speakers; (b) two discourse types: semi-spontaneous discussion vs. negotiation role play; and (c) two cultural contexts: intra- vs. intercultural context. Our investigation is couched within the framework of face (Goffman 1967) and politeness theory (Brown and Levinson 1987), which assumes the existence of Face Threatening Acts (FTAs), which human beings, in interaction, strive to mitigate. However, by contrast with the traditional view of mitigation, and following Briz (e.g. 1998, 2007), Hernández Flores (2002), Bravo (2003b) and Albelda (2005), among others, we assume that mitigation need not imply protection of the hearer‟s (H) face, i.e., it does not necessarily express politeness, but can just as well stem from the speaker‟s (S) own face concerns only. In our proposal, we therefore distinguish between allocentric mitigation (H and S face protection with politeness) and autocentric mitigation (S face protection without politeness). The empirical data is a selection of 34 conversations from Corpus AKSAM (Fant et alii 1996), split between four subcorpora and six speaker groups: (1) L1 Spanish speakers in intracultural discussions; (2) L1 Swedish speakers in intracultural discussions; (3a) L1 Spanish and (3b) L2 Spanish speakers in intercultural discussions; (4a) L1 Spanish and (4b) L2 Spanish speakers in intercultural negotiations. The material has a total duration of about 20 hours and comprises around 232 000 words. On the basis of previous work, including studies of Spanish and Swedish cultural values and face concerns (chapter 2), and a characterization of the different discourse types included in the corpus, assigning them certain contextual features (chapter 3), we present three quantitative hypotheses, according to which we expect to find a more frequent use of mitigating expressions: (1) in L2 than in L1 speakers of Spanish; (2) in negotiation than in discussion; and (3) in an intercultural context than in an intracultural context. In chapter 4, we investigate 28 different mitigating expressions in the shape of single words or word like units, called mitigation markers (or mitigators). These are divided into two main categories: (1) epistemic mitigators, and (2) approximative, generalizing and limitative mitigators (approximators). Through detailed qualitative analyses of contextualized corpus examples, we account for the mitigating functions of these expressions in each speaker group. Chapter 5 provides a deeper examination of a strategy that goes beyond the word unit, the so called concessive moves. These moves have two parts, conjoined by the adversative conjunction but („pero‟), out of which the first part mitigates the potentially face-threatening content of the second part. The concessions are divided into two major categories depending on their orientation in terms of face protection: allocentric and autocentric concessive moves. The qualitative analyses in chapters 4 and 5 are complemented with quantitative and statistic analyses (ANOVA), in order to accomplish the contrastive second aim of the study and test the hypotheses. Our findings show significant differences in a number of aspects, partially corroborating the hypotheses and partially pointing in other directions than the expected. On a general level, speaker group (L1/L2) as well as type of conversation (intra- and intercultural discussion and intercultural negotiation) are shown to have an influence on the use of both mitigation markers and concessive moves in our corpus.