friendofcars - Tumblr blog

friendofcars · 7 days

Text

Hello! Here is data on point of view distribution across characters in The Dreamer Trilogy (which I will abbreviate as TD3) as a follow up to my TRC data from last year (viewable here). A rather long-winded discussion of the data, methods notes, and some supplemental figures and tables are under the cut. As it was not possible to include all values and stats in this post (nor in the alt text for image IDs), my spreadsheet can be viewed by clicking here,

This project quantifies and visualizes the distribution of chapters and pages in the books of TD3 across characters from whose POVs the story is told. I didn’t have much of a hypothesis going into data collection/analysis, especially not like I did for the TRC data, but I did expect to see Ronan’s POV having the most chapters and pages for the entire series, given the fact that he is the most central of the protagonists. I don’t think page time is the be-all-end-all for a character’s importance, of course, but it is still interesting to consider how spending more time from certain perspectives affects the perceived narrative. I won’t get much into that aspect of analysis in this post, but if anybody would actually like to discuss that, I’d love to!

Results (and Interpretation):

TD3 consists of 173 chapters and 1184 pages (using the U.S. hardcover editions), making the average chapter 6.84 pages. The longest chapter is 16 pages, and the shortest is 1 page.

Figure 1A: The average chapter in Mister Impossible (MI) is considerably longer (9.26 pages) than the average chapter in Call Down the Hawk (CDTH) (6.00 pages) and in Greywaren (GW) (6.40 pages), which makes sense as MI has just 38 chapters while CDTH has 80 and GW has 55 (see Fig. 2). To me, the effect of the longer chapters (and therefore extended time with the current POV character) makes the various POVs feel more temporally distant from one another- not in a narratively incoherent way, but in a way that echoes the sense of isolation experienced by dreamers and weaponized by Bryde as he tries to convince Ronan and Hennessy to abandon their loved ones.

Figure 1B: Chapter length is fairly consistent amongst POVs across the series. Matthew has the longest average chapter length (8.40 page) over a small set of chapters (5 total)- his character development (as told from his own POV) is limited to a small number of instances, which may have stretched his chapters a bit longer. The 'Other' category has the shortest average chapter length (5.13 pages) (Fig. 1B); it includes the typically short chapters from witnesses of Visionary explosions/aftermath (Mags, Dabney) as well as Nathan's manifesto excerpts. (As a side note, I've described the chapters depicting memories from the Barns as 'Mór and Niall.' These chapters do not collectively portray an equal balance of their POVs, but this was the simplest way to categorize them.)

Figure 2A-B: These graphs are representations of chapter distribution across POVs in TD3 in terms of chapter count (2A) and proportion of total chapters (2B). Some observed trends include Declan's proportion of total chapters remaining quite constant throughout the series, Ronan's decreasing, Hennessy's proportion of chapters nearly doubling from CDTH to MI (and staying at a similar proportion to MI in GW), and Jordan's proportion following an opposite trend (consistent proportion in CDTH and MI, followed by a more than 50% drop in GW). Carmen's proportion of chapters also declines after CDTH.

Figure 2C: This graph compares total chapters per character POV over the entire series. We can see that the largest proportion of the series is told from Ronan's POV (53 chapters, or 0.306 of all chapters). To put that in perspective, Hennessy has the next highest number of chapters (26, or 0.150 of all chapters), which is just under half the number of Ronan's. If all characters had an equal number of chapters from their POV (including the miscellaneous POVs as one category labeled Other), they would each have 21.6 chapters, represented by the horizontal dashed line; Declan, Jordan, Carmen, and Hennessy all have chapter counts relatively close to this number.

Figure 2D-E: These are representations of page distribution across POVs in TD3 in terms of page count (2D) and proportion of total pages (2E). Trends are similar to those depicted in 2A-B, but 2E does make Declan's increased proportion of page time in GW salient.

Figure 2F: This graph compares total pages per character POV over the entire series. The dashed line shows that if each character (plus the Other category) had equal page time in the series, readers would spend 148 pages with each POV. Again, page data is similar to chapter data, but comparing graphs 2C and 2F gives a clear visual indicator that Jordan's chapters (on average, 8.11 pages) are longer than Carmen's (on average, 6.08 pages), since Carmen has visibly more chapters in 2C yet nearly the same number of pages as Jordan in 2F.

Figure 3: Figure 3 shows distribution of chapters (3A-B) and pages (3C-D) in CDTH, as well as average chapter length for each character POV (3E). An equal distribution of chapters would have been 13.3 per character, and an equal distribution of pages would have been 80.0 per character. The 'Other' category included chapters from the perspectives of Lock, Breck Myrtle, Shawna Wells, Jason Morgenthaler (and Lin Draper, briefly, in the same chapter), Mags Harmonhouse, and Dabney Pitts. Carmen's average chapter length in CDTH (4.67 pages) is the lowest single-book average for character POVs appearing throughout the entire series. (Nathan's average chapter length is just 1.00 [Supplemental Figure 2], yet his POV only appears in GW via his manifesto excerpts, and while I have attributed these chapters to his POV, I interpret the POV as actually ambiguous. As with Kavinsky's text in TDT, it's not absolutely certain if we are reading from the writer or the reader's perspective [although in TDT, due to the lack of Kavinsky POV elsewhere, it's probably the latter]).

Figure 4: Figure 4 shows distribution of chapters (4A-B) and pages (4C-D) in MI, as well as average chapter length for each character POV (4E). An equal distribution of chapters would have been 5.43 per character, and an equal distribution of pages would have been 50.3 per character. The 'Other' category included two chapters, both with what I deemed omniscient narration. Declan had the shortest chapters in MI (8.20 pages), and Jordan had the longest (11.4 pages, the longest average for a character for a single book in this series).

Figure 5: Figure 5 shows distribution of chapters (4A-B) and pages (4C-D) in GW, as well as average chapter length for each character POV (4E). An equal distribution of chapters would have been 6.88 per character, and an equal distribution of pages would have been 44.0 per character. The 'Other' category included Nathan's manifesto excerpts (3 chapters), 1 chapter from Liliana's POV, and 3 other chapters with omniscient narration. While Ronan never has the longest chapters, his chapters are shorter relative to other POVs in Greywaren, perhaps as a result of the way his chapters are written during his time asleep/in the sweetmetal sea. I have not yet investigated whether chapters tend to be longer while characters are awake vs asleep or dreaming, but that's something that could be measured from the existing data in the spreadsheet! There is also a dramatic drop in Jordan's POV time in GW compared to the previous two books, perhaps because of her increased divergence from Hennessy and desire to establish a life that follows her own narrative.

Other findings: A major difference I noted between TRC and TD3 was the lack of split chapters in TD3. In TRC, the data analysis was made slightly complicated by having to account for the fact that a non-negligible number of chapters would make a distinct and discrete switch between POVs partway through. While I did not observe this in TD3, I did encounter more ambiguous/nebulous POVs as I previously mentioned. The increased presence of omniscience in the trilogy, for me, contributed to the increased sense of scale and stakes compared to TRC. This increased continuity amongst POV (not amongst core/recurring POV characters, but amongst groups of characters/communities depicted in the omnisciently narrated chapters) also contributed to a sense of dissolution of barriers and identities, perhaps thematically in line with Ronan's character development and increasingly holistic perspective of both his humanity and otherworldliness (although Ronan is not necessarily featured in these 'boundary-breaking' chapters). I also briefly looked at occurrences of back-to-back chapters from the same POV; this happens most frequently for Ronan in all three books, mainly in CDTH, and sometimes featuring a dreaming chapter directly before an awake chapter (or vice versa) in immediate succession. Declan (MI), Carmen (CDTH), and Jordan (CDTH) all have a pair of back-to-back chapters at some point in the series; Hennessy has 2 (MI, GW).

Conclusions: In all honesty, despite this project being quite fun and fulfilling and of course, worth doing, I do not think I have any particularly insightful conclusions about the data beyond what I've already discussed. Ronan took up the largest share of the chapters and pages as expected, although I am not sure I expected this to be true by such a large margin. I also was surprised that Declan did not have more chapter/page time, but it is possible that his notable inclusion in chapters from other characters' POVs increases his prominence in the series (and I suppose this is probably true for all characters who frequently appear in chapters outside their perspective). As with TRC, the number of POVs expands as the series develops, often with the effect of increasing the scope of the story's implications, and perhaps, more importantly, showing the story from additional angles that contextualize and/or distort narrative established by other characters' perspectives. I hope you've enjoyed exploring the data as I have, and those interested in my methodology may continue reading below!

Methods:

Data collection was straightforward in the sense that I simply counted the pages in each chapter and then assigned each chapter to a character based on the POV represented. The POV character assignment was more difficult than it was for TRC, as TD3 has more omnisciently narrated chapters, which in itself is easy to categorize, but they often zoom in on or are 'biased' towards the experience of a particular character, so I had to make some decisions as to what, for me, constituted sufficient focus on a character’s internal narration and expression vs. omniscience. In the spreadsheet, I took notes on these more subjectively driven decisions. Again, you can view it here! It also contains data on whether the chapter is from an awake or dreaming POV, and has the first lines of each chapter, among which are some fun repeating patterns.

For bar graphs with dots, each dot represents a single chapter. You may also notice that the graphs are missing p-values from statistical tests this time around! This is because, since completing the TRC data, I’ve realized that such measures of uncertainty re: significant differences are not appropriate for my dataset, which is not a sample representing a population, but rather a complete group of chapters (so parametric tests are not necessarily helpful or valid). However, I still like to run the tests for my own amusement and to see what the results would be if this were a dataset for which ANOVA and contingency tests were appropriate, so I have standard deviation bars on the graphs where calculable (but no standard deviations in the text of the results section for legibility) as well as the p-values in tables at the end of this post for anyone also curious. I did still calculate the numbers of chapters and pages that would represent an equal distribution across POV characters, which are represented by the dashed lines on the relevant figures. I think this is helpful to visually gauge 'over-representation' and 'under-representation' of character POVs.

Below are the supplemental figures showing all character POVs rather than lumping some together in an 'other' category. The MI data in figure 4 is not expanded below because the chapters designated as 'other' were omniscient and thus would have remained in the same category.

And finally, here are the omitted p-values, if you'd like to pretend along with me that all the chapters in TD3 are not a complete set but rather a representative sample of a greater population of chapters that's out there in the universe. :) When I give a p-value below the 0.05 threshold but still write 'no significant differences amongst any combination of characters, I mean that the p-values generated for the comparisons between each possible pair of characters were all above 0.05, which are distinct from the overall p-value generated from the ANOVA.

#i was so awake and productive in early january when i was finishing this up 😭#anyway i am thinking about revising this analysis + the trc one from last year and posting it all on ao3 #i think this would be more legible there #(maybe? i think it would be on desktop at least)#i say that as if i have the energy to do so #anyway

102 notes · View notes