This case study reports on nested projects within a larger interdisciplinary project, utilizing linguistics and sociology, arising from a corpus of conversational data. Projects were data-generated, i.e. what happened between participants dictated the research questions. The data are recordings of parent-child interaction, and our viewings/hearings of these recordings. In one project, a recording device was given to the father, an English-speaker, who recorded a number of conversations with his bilingual son over an eighteen-month period. Researchers were not present during conversations. Recordings were transcribed according to the orthographic system developed by Gail Jefferson, and analysed using the research approach Conversation Analysis. Certain changes in the “systematics” of talk were observable over time due to the longitudinal aspect of recording the same participants over time, from when the child was 3 years old to when he was nearly 5. One of the lessons from this case is how little data are required from which to generate material for analysis. These ‘small’ data contain a multitude of interactional phenomena for explication.