The thesis begins with a conceptual model of the way that language diversity affects the strategies, organisation and subsidiary control policies of multinational companies. The model is based solely on the researcher's personal experience of working in a variety of international management roles, but in Chapter 2 a wide-ranging review of related academic literature finds evidence to support the key ideas. The model is developed as a series of propositions which are tested in a comparative case study, refined and then re-tested in a global survey of multinational subsidiaries. The principal findings of the empirical phases of the thesis endorse the main tenets of the model: (i) That language difference between parent and subsidiary will impair communication, create mistrust and impede relationship development; (ii) That subsequently the feelings of uncertainty, suspicion and mistrust will influence the decisions taken by the parent company; (iii) They will have heightened sensitivity to language issues and will implement policies to manage language differences; (iv) They will adopt low-risk strategies in host countries where they are concerned about language difference; (v) They will use organisational and manpower strategies to minimise the consequences and risks of the communications problems with the subsidiary; (vi) As a consequence the level of integration and knowledge flow between parent and subsidiary will be curtailed; (vii) They will adopt styles of control that depend least on their ability to communicate with their subsidiary. Although there is adequate support for all of the above conclusions, on some key points the evidence of the Case Studies and Survey is contradictory. The thesis, therefore, closes with an agenda for further research that would address these inconsistencies.