Using corpus techniques to evaluate gender representation in international development reports

The Governance and Social Development Research Centre (GSDRC) at the University of Birmingham prepares research reports for the Department for International Development (DfID). These reports summarise the relevant literature in response to questions that have been posed by DfID. When writing these reports, GSDRC researchers are asked by DfID to consider gender in their evaluations of the effects of any intervention.

On 24 July 2013, the GSDRC team held an Away Day at which they discussed issues relating to gender equality, and they asked us to contribute to this event with a report of how corpus tools can be used to evaluate gender representation in their reports. In advance of the meeting, GSDRC supplied us with raw text files of 100 recent reports that they had written and we then used Wordsmith Tools to derive frequency and concordance data from this international development report corpus. We focused on looking at words such as woman, girl, man, and other gender specific nouns, and on the contexts in which the words were used.

The main findings were:

  1. Women are mentioned significantly frequently in the reports (much more so than would normally be expected in a corpus of general English language texts).
  2. Women are mentioned less frequently in reports which:

    a) focus on macro-level issues relating to finance, aid and so on.
    b) foreground other social distinctions, such as disability.
  3. Women, probably like other social groups in the reports, are not discussed in ways that suggest they have much agency.

The response from the GSDRC team was very positive. We expect to follow this up by writing a blog post that presents details of the analysis, running a workshop for GSDRC staff that trains them in how to use corpus analysis tools (the freeware AntConc programme) and carrying out a further study with a larger corpus of international development reports.

Caroline Tagg

Paul Thompson