23 February 2011

The rise and fall of development fashions

This week is "fashion week" in London, where all the major designers get together to strut their stuff (think major development conference with more stylish clothes).

So what better time to preview the new Google Books Ngram Viewer or what Martin Ravallion, Head of Research at the World Bank, calls a "linguistic window on public awareness"?

A word is an I-gram, and so an N-gram is a phrase made up of I-grams. Michel et. al. 2010 (in the journal Science) have developed a body of N-grams based on 5.2 million digitised books from the period 1500-2008. The Viewer searches through the 500 billion words and counts the number of times the word or phrase appears and reports it, normalised by the the total number of words in the digitized set for that year.

So, if you wanted to search for key words and phrases over the past 500 years, you can do it very easily (in several languages--my examples below are in English)

See this set of 3 incidence lines for hunger, famine, malnutrition over the 1500-2000 period. Hunger and Famine have always been with us, but Malnutrition is a relatively new term, only coming into common usage in the 20th Century.

So, let's put some development fashions on the runway.

Agriculture: rural development vs sustainable livelihoods for the 1950-2008 period? Peak in rural development around 1982, steep decline thereafter, with a rise in sustainable livelihood, but not nearly at the same scale as the drop in rural development

Gender: women and development vs women in development vs gender and development vs women's empowerment. The first phrase peaks in 1992 or so, the second a year or two later, the third in 2000 and the last, women's empowerment, really took off in the late 80s and is yet to peak.

Climate: global warming vs climate change. Both terms take off in 1985, but global warming levels off in 1995 while the use of climate change powers on.

UN Agencies: for UNICEF vs FAO, see a tale of two agencies. In the late 80s UNICEF overtook FAO in book mentions and powered on, having been level pegging for about 30 years.

Bretton Woods: Bretton Woods vs Washington Consensus. The use of Bretton Woods has been declining since 1995 and the use of the Washington Consensus, although lower in incidence, is still on the rise.

Development: sustainable development vs wellbeing. Wellbeing has been with us for a while and is rising, while sustainable development peaked in 2000.

MDGs. MDGs took off in 2001 and are still on the rise.

Crises: world food crisis vs oil price crisis. massive peak for the first in late 70s and more modest peak for latter in late 80s.

Human rights and social justice: the former takes off in 1975, and the latter is steadily rising.

On accountability, transparency and corruption: corruption was written about much more in 19th century compared to the 20th, at least in English language books. Accountability and transparency are slowly catching up in the 21st century.

Charities and NGOs: charities peaked in the mid 19th century and NGOs are a 1980s phenomenon.

Social protection, social capital, safety nets. Interestingly social protection has been used more widely than safety nets for a very long time. social capital dwarfs both, however, taking off around the time of Putnam's book on Italy.

As you can see, one can go on for ever (it's addictive).

Its not quite clear how it can best be used for serious research, but perhaps the historians are already using it (it was launched in December 2010).

As Ravallion points out the Viewer does have biases, including:

* is the 5.2 million book sample from Michel et. al. 2010 representative of the 15 million digitized books from the 40 university libraries all over the world?

* are the books representative of thinking at the time or just the current thinking of literate people?

The potential for misinterpretation of phrases and words is nontrivial, but as a starting point this tool might help us look backwards and learn from the past, give issues more of a historical context and remind us of how faddish we can be in development (what is the development equivalent of platform shoes?).

In helping us look backwards the Viewer might even help us look forwards a bit better.

Let us know about any interesting graphs you made with the Viewer.

No comments: