An international school (BES) in Greece, overwhelmingly attended by Greek origin children, has adopted, as its language policy, English as the ‘official’ medium of interaction, including in the Reception classroom, the target of this research. That is, through its language policy, the school aims to promote the learning and use of English throughout school. At the same time, the school has adopted ‘free interaction’ in designated play areas as its pedagogical approach. The aim of this approach is to promote learners’ autonomy and, in the particular case, it could be interpreted as including the possibility of using Greek. Thus, a conflicting situation has developed: how to reconcile the school’s English monolingual language policy and the pedagogical approach in the play areas? Reception educators are expected to police the use of English in the kids’ play areas without however undermining children’s autonomy and/or disrupting their ‘free interaction’. The feelings and views expressed by educators show that they are seriously concerned about how this conflicting situation can be approached. The aim of this thesis is to respond to this issue of concern by providing a detailed description of how the school’s conflicting policies are actually lived in the educators’ and pupils’ language choice practices in the play areas of their classroom. By adopting the Applied Conversation Analytic perspective of “description-informed action” (Richards 2005), a perspective whereby practitioners are made aware of their own practices and are left to “make (their own) decisions regarding the continuation or modification” of their own policies and practices (Heap, 1990: 47), the aim is to raise BES stakeholders’ awareness about the possible advantages, possibilities and limitations of their policies and practices in Reception, and thus pave the way to more informed language policy making and practice in the school. The data consists of audio-recorded naturally occurring child-child and childadult interactions in the school’s play areas. The analytic framework draws on Spolsky (2004), for whom “the real language policy of a community” resides in its language practices (hence the notion of ‘practiced language policy’), and on conversation analytic methodologies applied to language choice (Auer 1984, Gafaranga 2000, 2001, 2005, 2007a, 2009). The key finding is that, adult school members and children respond to the school’s conflicting policy demands in different ways, i.e. by orienting to different ‘practiced language policies’. On the one hand, as the adults’ ‘medium request’ (Gafaranga 2010) practices in the kids’ play areas demonstrate, from the adult perspective, at all times, participants need to attend to a language preference that is ‘institutionally-assigned’, i.e. adults orient to a ‘practiced language policy’ that is in line with the “declared” (Shohamy 2006) English monolingual language policy of the school. This shows that they have responded to the school’s conflicting policy demands by prioritising the school’s language policy (use of English) at the expense of the pedagogical approach (learners’ autonomy). On the other hand, children approach the conflicting situation differently. Children seem to have developed an alternative ‘practiced language policy’ according to which language choice during peer group interaction is not organised around the school’s “declared” (ibid) language policy but around their interlocutor’s “linguistic identity” (Gafaranga 2001). This alternative language policy allows the kids to attend to the pedagogical approach (learner autonomy and free interaction).