Imagine you see a sign on a door that says “Hazardous Area Do Not Enter” and your friend is already turning the door handle and about to walk in. What do you say to your friend? According to rational approaches to communication, language users seek to optimize their production and comprehension of language in a way that allows communicative goals to be accomplished an effective way. How language production and comprehension unfold are subject to (i) situation-specific properties and real-world knowledge, (ii) constraints on human cognition and our information-processing abilities, and, given that languages differ in their grammatical properties (e.g. pronominal paradigms), to (iii) crosslinguistic differences, among other considerations.

I will present a series of psycholinguistic experiments that explore these themes and their relation to the ideas of rational communication, predictability and surprisal. The studies focus on different case studies, including (a) reference (production and interpretation of different pronominal and demonstrative forms), (b) descriptions of part-whole relations, as well as (c) causal relations (cause-effect sequences on warning signs vs. narrative stories). In addition to new insights into (lack of) simple predictability effects on linguistic form, our findings provide evidence for situated language use: We show that the communicative goals of different contexts can, for example, reverse speakers’ word order choices. Furthermore, we provide new evidence for a ‘just right’ approach to information processing, suggesting both highly typical and highly atypical configurations can be harder to process than those at a medium, ‘just right’ level of informativity. Put together, these studies shed light on how linguistic form, non-linguistic situational factors, and cognitive constraints interact to make language a highly context-dependent, dynamic system.