In the 1920s, Bronislaw Malinowski – in the tradition of Herder and Humboldt and based on his experience during his field research on the Trobriand Islands – pointed out that language is not only an instrument of thought, but first and foremost a tool for creating social bonds and accountability relations in more or less ritualized forms of social interaction. Language is a mode of behavior and the meaning of an utterance is constituted by its pragmatic function: it can only be understood in relation to the context in which it is embedded. The rules that guide communicative behavior vary immensely in different cultures and have to be learned to achieve communicative competence within a specific speech community. This learning results in the understanding of how the speakers structure, pattern and regulate their ways of speaking. Malinowski’s ideas had an increasing impact in anthropology and linguistics – especially in pragmatics – and led to the formation of the subdiscipline “anthropological linguistics”. This paper presents three observations of the Trobriand Islanders’ attitude to their language Kilivila and their actual language use in social interactions which I made during my fieldwork on the Trobriand Islands. They illustrate that whoever wants to research the role of language, culture and cognition in social interaction – be it linguist or anthropologist – must know how the researched society constructs its reality. Researchers need to be on ‘common ground’ with the researched communities, and this common ground knowledge is the indispensable prerequisite for any successful research on language, culture and cognition manifest in social interaction.