Network
[ work in progress ]
In typical Pynchon fashion, paranoia pervades Gravity's Rainbow: Slothrop's Puritan reflex seems to infect everything that he touches, turning pure coincidences into conspitorial structures. Nothing unusual, though, since Pavlovian psychology has long ago taught us that associations are the backbones of human understanding and memory. Such is the way that the novel expands around associations, conditioning, drawing seemingly far-fetched (who knows?) connections between our little paranoia agents, who all oscillate between the fear that nothing relates and the suspicion that everything does. So, what better way to analyse Gravity's Rainbow than looking at these networks between characters, names, and words?
This, of course, has been done countless times before: Richard Poirier already traced these networks in his 1973 review; Luc Herman, Robert Hogenraad, and Wim van Mierlo (2003) measured chapter-level word covariation; Martin Paul Eve (2015) mapped frequent terms and the connections between them; David Letzler (2016) combined social-network analysis with word frequency, keyness, and topic modelling.
On this page, therefore, I provide my own attempt to present the novel's various webs in interactive form.
Co-occurrence
'They are in love. Fuck the war.' (1.06). The lovebirds Roger and Jessica share only eleven episodes compared to Katje and Slothrop's sixteen, the latter obviously leading by count; then why do we have so much more attachment to the former's relationship? In this matrix we don't simply count co-occurrence episodes: it's much less surprising to find two characters present together in a crowded episode, compared to when a pair of characters dominate their own episodes. The resulting measure is 'prominence-corrected co-occurrence': reach 1.6, the highest weight in the matrix, whereas have only 0.9. Amongst the top forty pairs we see four out of the five communities identified in the character Atlas above. form London's strongest tie, Rocket's; appears in seven of the ten strongest cross-community pairs. This result concurs with prior analysis by David Letzler (2016), which identifies Slothrop with Tantivy and Roger with Jessica as strongest. Martin Paul Eve (2015) groups Roger, Jessica, and Pointsman, using frequent-term proximity across the whole text rather than episode co-occurrence. The methods are here.