This paper examines the Spanish form o sea ‘that is’ within a discourse model based on the notion of coherence relations. The advantage of this discourse framework over others is that it allows for an independent definition of Discourse Markers (DM) as a linguistic category.

Some of the main theoretical approaches to the identification of DMs try to characterize them by a set of features that they are supposed to share, however, neither of these features alone or in combination, are able to provide a full characterization of a category of DMs (Jucker and Ziv, 1998). The main problem lies on the fact that only a series of a priori selected groups of expressions constitute the empirical basis of these studies, therefore the definition of DMs is entirely dependent on the particular set of forms chosen to be analyzed and does not always apply to other possible members of the category. A different perspective to the definition of DMs is the analysis of these expressions as part of an independently motivated discourse framework. This is the approach adopted in our work. Following premises of models of Rhetorical/Coherence Structure Theories (Mann and Thompson, 1988; Sanders et al., 1992; Hobbs, 1990;1995), we understand discourse as a series of units connected to one another in a coherent manner. The different types of connections which provide coherence in discourse are called coherence relations. Discourse is structured in interrelated components in which specific types of relations become more relevant (Redeker, 1990; 1991). The semantic component is the one in which the relations between the propositional contents of discourse segments provide coherence. The pragmatic component is the one in which relations expressing the speaker’s attitudes and beliefs towards the reality of the states or events expressed in the discourse segments are responsible for establishing coherence. And the textual component is the one in which relations representing sequential transitions are most relevant. Coherence relations are also the key concept in the definition of DMs. We understand DMs as discourse functional units whose primary role is to make coherence relations explicit; DMs have the specific function of making evident to the listener a particular coherence relation between two discourse segments. Integrating DMs in an independent discourse framework on the basis of their functions overcomes the a prioristic methodological problems mentioned above.

On the basis of these tenets, we set out an empirical study on a corpus of 16 hours of natural conversations in Peninsular Spanish extracted from the CREA (Corpus of Reference of Modern Spanish). The results obtained show that o sea is indeed a DM in Spanish, since it is functionally capable of representing semantic, pragmatic and textual relations. In quantitative terms, o sea expresses primarily semantic relations between units in discourse, and then to a lesser extent pragmatic and textual relations. The semantic relations represented by o sea all belong to the general category of relations of restatement, that is, the unit introduced by this DM expresses a re-elaboration of the propositional content of the previous unit. The re-elaboration can take the form of neutral restatement, elaboration specific, or correction. In the pragmatic domain, on the other hand, o sea represents the relation of conclusion: the unit introduced by this form expresses a reasonable conclusion that the speaker derives from the information presented in the first member of the relation. Finally, o sea establishes also the textual relation of restart: the speaker, having introduced the first part of a discourse segment, feels the need to redirect her speech into a different direction.

The theoretical importance of these results is twofold. First, they argue against assumptions that different types of relations are represented by different DMs (Fraser, 1999). Second, they allow us to view the traditional proposals of DMs within a wider perspective. A DM is no longer a pragmatic operator exclusively, but a discourse functional unit capable of representing pragmatic, semantic and textual relations.