The language of formative assessment, sometimes referred to as "assessment for learning" (AFL), is becoming ubiquitous within international educational discourse. Despite this, levels of common understanding might be considered questionable since the concept appears to have been subject to many different interpretations and to have resulted in varying implementations. The reason for this ambiguity partly lies in the way that the language of formative assessment or assessment "for" learning has increasingly been applied beyond the context of classroom practice; shifting the focus of assessment from explaining how classroom learning interactions can enhance individual development to considerations involving larger-scale, system-wide accountability purposes. This shift appears to contradict the original spirit of AFL and has led to a blurring of the formative and summative boundaries of assessment. This article presents some empirical data from a small-scale research project to suggest that this issue has a practical impact in the classroom. The article considers the variety of ways that the language of AFL is used by teachers in differing international contexts and suggests that a conflict of purpose might underlie such variations. The article goes on to suggest the need for further research addressing some of the fundamental paradoxes within assessment discourse caused by this conflation of purposes, and identifies areas for further research and investigation.