The authors have done what
is the best job I’ve seen so far of explaining what topic modeling does. There is some fine print that
relates to my concerns over topic modeling. In general I find topic modeling only useful for a very general sense of content, but not very much "aboutness" which is to say how the "topics" appear semantically.
However, the cocitation map is a really interesting way to understand how authors relate to one another, (again you need to read the explanation fully to understand what the network reveals).
My own network analyses looks at conferences not
publications but I was curious about how some of the major characters in the
Politics of Women’s Culture fared in this analysis Signs. However, the cocitation map is a really interesting way to understand how authors relate to one another, (again you need to read the explanation fully to understand what the network reveals).
By the cocitation map Adrienne Rich appears fairly negligible, as compared to say Foucault with whom she is
surprisingly connected, albeit rather tenuously, (in comparison she shares a stronger affinity with Carroll
Smith-Rosenberg)
static Rich Cocitation |
Static Foucault Cocitation |
This seemed low for someone like Rich so I began to think about other ways to measure presence in Signs. According to JSTOR data, her most influential article, Compulsory Heterosexuality and Lesbian Existence, has been accessed 7669 times in the past three years!
I started simply by searching for"Adrienne Rich" in JSTOR's Signs Archive. Her name appears in 219 items(as near as I can tell there are 3,858 individual items from Signs in the database. The analysis in Signs@40 is based on 1800+ articles. Omitting back matter (ads) gets us 179 individual items and 159 author-attributed items (removes editorials). I looked at three variables, authors who invoked her name at least once, dates of issues that contained at least on mention of her by name, and by type of item in which her name appeared. By author the results were pretty much as I’d expected. Only 9 authors had cited her more than once, with Susan Gubar coming in at 3.
By dates of issues that contained at least one mention of Adrienne Rich things started to get interesting. The peak of Rich’s popularity is the Summer of 1984 "The Lesbian Issue" when she appears in 12 separate items. In total “Adrienne Rich” appears in 81 of the roughly 160 issues of Signs HALF of them!! While she tapers off in the 21st century Rich has been a figure on the pages of Signs for 35 of its 40 years. (from 1975-1998 she appeared at least once yearly, missing 1999, 2002, 2004, 2007, 2013)
dates of Signs Issues with at least 1 "Adrienne Rich" |
The majority of the mentions of "Adrienne Rich" are in research articles, pointing to her central role in feminist discourse.
items from JSTOR metadata by TYPE |
The analysis done for Signs@40 reveals that Rich’s Compulsory Heterosexuality and Lesbian Existence contributes to a wide range of “topics,” but that doesn’t really get at how authors wrote about Rich. For that we need different tools, such as computational linguistics.
I am also curious how Adrienne Rich comes to be so relatively isolated in the cocitation network given her frequency in Signs. Is it that the authors who write about "Adrienne Rich" are less likely to footnote overall? Is the lack of cocitation representative of the incredibly broad range of topics into which Rich is brought to bear, but perhaps in conjunction with less frequently cited authors? I'm also curious about how RIch's two signal concepts "of woman born" and "compulsory heterosexuality" appear in Signs. My expectation is that semantically "of woman born" will diachronically tip negative, while "compulsory heterosexuality" will remain positive, but that remains to be seen once I do the linguistic analysis.
I am also curious about other authors, like Audre Lorde. The Signs@40 digital projects are going to provide a great context for the computational linguistics analysis of a broad range of feminist periodicals that I'm analyzing for the upcoming AHA.
No comments:
Post a Comment