Pragmatics, the study of language use in context, has largely been positioned as an autonomous branch of linguistics that has little or nothing to do with semantics. However, it seems counterintuitive to me to attempt to understand language use without at the same time trying to understand what it means. Catherine Travis's publication is in this sense an important piece of work; it showcases how language use may be more fruitfully studied through focusing on language meaning. The aim of Travis's book is to identify the “conversational conditions” under which a set of discourse markers is used in Colombian Spanish, and, on the basis of those “conversational conditions,” to “determine and explicate the meanings of these markers” (2). In particular, it attempts to “demonstrate that the pragmatics of use of the discourse markers under consideration is semantically driven: the use of discourse markers is determined by their inherent meanings, which interacts with context-driven features to give rise to different pragmatic functions” (2).

The Colombian Spanish discourse markers that Travis selects for study are bueno ‘well, all right, OK, anyway’, o sea ‘I mean, rather, that is to say’, entonces ‘so, then’, and pues ‘well, so, then’ (2). For each discourse marker, she first tries to establish the relationship between the marker and the non-marker form (e.g., discourse marker bueno and adjective bueno). Then, looking at the range of functions that each discourse marker performs, she proceeds to try to find out whether “homonymy, polysemy, or generality of meaning” is involved (4). More specifically, for each of these discourse markers, Travis traces its semantic development, looks at its functions, works out the number of related meanings it represents, and describes each of these meanings in the form of a paraphrase. She proposes “four different meanings” (78) for bueno, “three related meanings” for o sea (133), “three core meanings” for entonces (172), and “two-way” polysemy for pues (240).

I find Travis's book highly commendable for a number of reasons. First, it shows us that, contrary to what some linguists might think, discourse markers are not void of semantic content. It fact, it has meaning that can be clearly stated from the insider's perspective. Second, it gives us a good representation of what happens during semantic change. In a new phase of development, some parts of the meaning remain. At the same time, some other parts may be lost and some new parts may be gained. Third, it opens up a number of interesting and important issues for discussion (e.g., what goes into meaning, how meaning is related to functions, how to identify polysemy in relation to discourse markers, how to construct the path of semantic change). Fourth, it is an easy read, for Travis seems consciously to avoid the unnecessary use of technical labels. Last but not least, its interdisciplinary nature would give it wide appeal to scholars working in a number of fields: semantics, pragmatics, discourse analysis, language education, sociolinguistics, anthropological linguistics, historical linguistics, and Romance languages.