ABSTRACTInteractions between primary care‐givers and people with mental handicaps have been a focus of eco‐behavioural studies. Many of these have classified verbalisations by staff and the people resident in long‐stay settings to indicate the nature of the linguistic environment and show that there is a low level of high quality interaction in these settings. A comparison between two separate wards in the same hospital was carried out to obtain baselines as part of a study of the efficacy of staff training procedures. The wards were identified as being very different from each other in management practices, level of communication ability, and degree of support required by the people living in them.The present study examined the use of an observational schedule which yielded data for fine‐grained analysis of the patterns of interaction between residents and staff; according to population density and by staff grade. The results of the baseline measures indicated very few differences in interaction patterns between the wards despite other differences in régime. While much previous research has suggested that different management approaches, such as staff training, are important factors in setting differences, it is only in‐depth investigations which examine these aspects in terms of the patterns of language use and activity that highlight the extent of the effects of institutionalisation on staff and the people in their care.The implications are that training needs to be targeted at communication to overcome some of the features of existing interaction patterns. Replacing large institutions by smaller facilities, where changes in management or physical features only replicate the activity, staffing, and systemic patterns of large hospitals, could lead to very little fundamental change in interaction patterns for the people who live in them unless effort is directed towards facilitating their development of socially interact