The concept of classification has been introduced to mathematics education by Dowling (1996). In subsequent years it has proved to be very useful for research in different fields such as for analyses of mathematics school books (Dowling 1998), teacher discourse (Morgan, Tsatsaroni & Lerman 2002), teacher education practice (Ensor 2004), and classroom interaction (Gellert & Hümmer 2008, Hoadley 2007). Dowling (1996, 1998) has extracted the concept of classification from Basil Bernstein’s theory of pedagogic codes (1975, 1990, 1996). Bernstein (1990, 1996) is interested in descriptions and changes of instructional and regulative classroom rules, that is in modifications of what knowledge is accessed in classrooms and of how this knowledge is made accessible. He defines classification as: “We have said that dominant power relations establish boundaries, that is, relationships between boundaries, relationships between categories. The concept to translate power at the level of the individual must deal with relationships between boundaries and the category representations of these boundaries. I am going to use the concept of classification to examine relations between categories, whether these categories are between agencies, between agents, between discourses, between practices.” (Bernstein, 1996, pp. 19-20) In the context of school mathematical knowledge, classification may refer to categorizing areas of knowledge within the mathematics curriculum. Strong internal classification means that clear boundaries between mathematical areas are maintained. Strong external classification indicates that few connections are made to other disciplines or everyday practice. In order to analyse school mathematics discourse in streamed pedagogic practices it is useful to develop a multidimensional conception of how classification is realized in mathematics classrooms. I distinguish three dimensions for the purpose of analysis. The first dimension connects with Dowling (1998) who uses two dimensions of classification to analyse school mathematical activities as initiated by textbook tasks: content and expression. The content dimension draws on the strength of the boundary between mathematical/arcane and everyday/mundane knowledge. This dimension seems to be useful for analyses of classroom interaction. As my interest is in classroom talk, the second dimension, graphical and textual characteristics, is less relevant. Instead, I distinguish forms of talk (register, sentence construction, reference …) as the second dimension of classification. A further dimension is necessary because the dimensions of content and forms of talk do not cover the complexity of students’ mathematical activities. In the third dimension I refer to the concept of praxeologies, which distinguish the praxis and the logos of school mathematical activities.

My research question are (1) How do streamed pedagogic practices in the early secondary grades differ in terms of their classification values? (2) What is the effect of the streamed pedagogic practices on students’ access to mathematical knowledge? (3) What, in the end, is school mathematical knowledge really in the different streams? I will draw on videotaped classroom data from Germany and Canada (Knipping, Reid, Gellert & Jablonka 2007), which are analysed by key incident analysis (e.g. Sturm 2000), to illustrate the research results.

The first findings suggest, that there are differences in all of the three classification dimensions that particulary disadvantage children in the lower streams. The issue of access to valued forms of mathematical knowledge is highly problematic in these classrooms.