Although I’m pretty committed to working with antconc (lesson 1 and lesson 2), I couldn’t resist another run at topic modeling to see what differences emerged.  All that led to some good debate via the twitters, which I storified and link to at end of this post

Using the full text OCR files sent to me by the kind people at JSTOR that I’ve been using with AntConc,  I ran 1977 through 1981 Off Our Backs through mallet to see what sorts of topics emerged.  I’d done that in the past, but without the full texts.

List of Topics

1. women children health child care disabled medical control birth woman
2. women gay lesbian st call center box write washington lesbians
3. rape prison case police court state defense violence judge trial
4. oob issue women backs collective de paper friends news articles
5. workers union rights women national government state federal labor public
6. music book feminist books lesbian woman order year poetry write
7. black men male woman lesbian lesbians sex book sexual power
8. abortion anti people life rights pro country tion woman human
9. women feminist movement work group feminists political conference men groups
10. people time don work years back woman didn ve home

so without looking into files that gave rise to the above some topics seem clear
1 medical
2LGBT community in DC which makes sense given that OOB is published out of DC,
3 something to do with rape trial
4 the process of publishing OOB
5 government policies about work?
6 different kinds of women’s “culture” (note lesbian in there)
7 hmmmmm
8 abortion rights
9 stuff about the movement
10 hmmmmm

I tweeted about this, and got a response from the every helpful Jon Goodwin

  1.  I’m nly vaguely convinced that topic modeling would tell me what I wanted, less so after reading about CDA today
Jon suggested that I expand my topics given the size of my corpus 2900+ items, so I ran more
right away I can see the judicial 1 has separated from rape 18
collective seems to stay the same, as does culture 8, medical 4, movement 14
new about the family 10 , clearer one about rape, and even clearer one about racism 13 , movement conferences  has emerged 15, relationships 16
pretty baffled by 2, 5
Jon Goodwin kindly  made a 30-topic browser of Off Our Backs (1978-1981).  using MALLET default stopword list, and I also turned on optimize-interval to weigh the topics. Threshold is .15. he browser’s text is clickable. main –> topic page w/ articles –>articles w/ assoc. topics&graphs (and JSTOR link).