Which textual features are difficult when reading and solving mathematics tasks?

International audience; Despite the digital revolution much of the mathematics practiced in schools is still tightly bound to two-dimensional texts.

This emphasis on text is neither surprising, nor inadequate, since mathematics has developed through a long history with the use of written text, consisting of natural language, mathematical notation and images.

Natural language is our native language consisting of letters and words (see e.g., www.oed.com).

Different features of the mathematics text are also important in written tests, since reading the text is part of the assessment.

If the text is hard to read, that difficulty can be relevant as part of assessing the communicative competence in mathematics.

Crucial is, however, whether potentially difficult textual features are part of what the assessment aims at.

This issue is investigated in the current study, using a synthesis of statistical results and qualitative analyses of task text.

A critical question is where to draw the line between necessary and unnecessary reading demand and how to judge which textual features are irrelevant and therefore should be avoided in mathematics assessments.

In the current study this aspect of reading demand is addressed through a small meta-analysis of four studies where different textual aspects in task text are analyzed in relation to task difficulty and task reading demand.

The theoretical starting point for the current research is an understanding of language as an essential part of mathematics.

It has been argued theoretically that the understanding of a mathematical object develops as the student develops her or his discourse on that object (see e.g., Sfard, 2008).

An understanding of mathematics discourse as part of what mathematics is, is in line with the theoretical interpretation of the statistical measure for demand on reading ability (DRA) used in the studies included in the meta-analysis conducted in the current study.

DRA is a measure of the unnecessary reading demand in a mathematics task, and within this interpretation lays also an assumption of a kind of reading demand that is relevant in mathematics tasks (see also Dyrvold, Bergqvist, & Österholm, 2015).

The purpose of the study is to contribute to the knowledge about which textual features in tasks are demanding and whether that difficulty is a mathematics relevant difficulty.

The research questions are: i) what conclusions can be drawn regarding reading demand in mathematics tasks in relation to textual features?

, and ii) how can the conclusions based on statistical analyses be interpreted in relation to a qualitative analysis of mathematics task text with a high reading demand?

The study consists of a meta-analysis and a qualitative analysis of tasks that stand out in the quantitative analysis.

Only four studies are included in the meta-analysis but even such a small meta-analysis do contribute to the development of knowledge since the analysis enables conclusions to be drawn that would not be possible to draw without such an analysis.

The qualitative analysis has a systemic functional perspective (Halliday & Matthiessen, 2014) and includes also images and mathematical notation.