We use language to construct meaning. But what exactly do we construct, how, and for whom? These questions are often formulated as a choice between conceiving of meaning as involving something in the world versus something in the head. In this talk I argue that not only is this a false dichotomy, but more importantly, we will not understand human language until we take both views on meaning equally seriously and place them both on equal footing. One important part of language – which formal semanticists have been most concerned with – specifically supports connecting our words to the things we refer to (be they real or hypothetical). Another important part of language – more the focus of cognitive linguistic research – helps us structure and categorize the world around us. The evidence suggests that these parts are inextricably intertwined, even though we often practice linguistics as if they can be isolated from each other, setting aside one or the other for convenience or out of conviction.

I will make my case by discussing some challenging data on which both formal and cognitive semanticists are typically silent. I’ll advocate bringing these data clearly into the empirical purview of a theory of meaning, suggest some strategies for handling them, and highlight perhaps surprising connections to certain proposals in morphological and syntactic theory. Finally, I’ll close by advocating a broader semiotic perspective on language, oriented towards helping linguists clarify the enterprise of linguistics in relation to cognitive neuroscience, on the one hand, and natural language processing and artificial intelligence, on the other.