All of the analysis so far has been done by year. But sometimes it is helpful to look at a number of years at once. In this chapter I do this twice over.
First, I’m separating the times in the data set into five eras. My idea was to have eras that correspond to natural time periods in philosophy, but which are also roughly equal in terms of the number of articles in each era. This suggested the following five-way division.
|Era||First Year||Last Year||Number of Articles|
Intuitively, I’m thinking of the eras the following way.
- Era 1 covers everything from idealism though positivism and World War 2.
- Era 2 is the postwar period, i.e., ordinary language philosophy in Britain and Quinean postpositivism in America.
- Era 3 is where the classic works of contemporary analytic philosophy were written by writers such as Kripke, Lewis, Putnam, Rawls, Thomson, Singer and Frankfurt.
- Era 4 is the one I have the hardest time conceptualizing, as it seems to me something of a period of consolidation between the revolutionary developments of the 1970s, and the series of new debates that open up in the 2000s.
- Era 5 is the first part of the current century, and is dominatedby a number of distinctive topics, such as reasons, vagueness, contextualism and Williamsonian epistemology and metaphysics.
In the second half of the chapter, I look at what happens when we break the data down by decades. This has the advantage of corresponding to time periods that people naturally understand, but two disadvantages. One is that data gets lost from either end of the data set. Before 1890 there are too few articles to do a useful analysis, and 2010–2013 isn’t a complete decade. The other is that the number of articles is very uneven across the decades. But despite these disadvantages, I think it’s helpful to get a decade-by-decade view at what was happening in the philosophy journals.
Within each of these parts, I’m going to run three studies. First, I’ll look at the distribution of the ninety topics across the eras and decades. Second, I’ll look at the distribution of the twelve categories across the eras and decades. And third, I’ll look at the distribution of individual words across the eras and decades. The latter is useful because it doesn’t involve processing everything that happened through a black box of an LDA implementation; we can just see what words were and weren’t being used. As well as being interesting in its own right, this helps us check the plausibility of what the LDA comes up with.