Language Log

« previous post | next post »

I downloaded rev.com's transcript of last night's vice-presidential debate, and did a bit of analysis — the most interesting stuff will come later, but to start with I did a couple of my standard simple-minded analyses, starting with the type-token plots:

It's somewhat interesting that Walz and Harris are so similar, and that Vance is kind of splitting the difference towards Trump's low lexical diversity (due to repetitive rhetoric).

I also calculated Vance and Walz's most characteristic words (or at least the words most differently used in this debate), using the method described and exemplified here.

Vance's top ten were

country 44 (5156.45) 6 (707.631) 50 (2939.1) 3.065
american 44 (5156.45) 7 (825.569) 51 (2997.88) 2.957
actually 31 (3632.95) 2 (235.877) 33 (1939.81) 2.866
tim 25 (2929.8) 0 (0) 25 (1469.55) 2.825
margaret 22 (2578.23) 0 (0) 22 (1293.2) 2.650
of 232 (27188.6) 146 (17219) 378 (22219.6) 2.573
lot 44 (5156.45) 13 (1533.2) 57 (3350.58) 2.352
policies 17 (1992.27) 0 (0) 17 (999.295) 2.329
walz 15 (1757.88) 0 (0) 15 (881.731) 2.187
illegal 14 (1640.69) 0 (0) 14 (822.949) 2.113

where the 8 fields in each line are:

1. Word
2. Vance's count
3. (Vance's count per million)
4. Walz's count
5. (Walz's count per million)
6. Summed count
7. (Summed count per million)
8. Estimated log odds that it's from Vance

(Sorry for all that, it's what my program emits and I don't have time now to fix it…)

Walz's top ten words were

there 14 (1640.69) 69 (8137.75) 83 (4878.91) -3.489
this 71 (8320.64) 154 (18162.5) 225 (13226) -3.257
minnesota 1 (117.192) 26 (3066.4) 27 (1587.11) -2.744
it 96 (11250.4) 169 (19931.6) 265 (15577.2) -2.656
senator 1 (117.192) 20 (2358.77) 21 (1234.42) -2.366
folks 2 (234.384) 21 (2476.71) 23 (1351.99) -2.269
sure 2 (234.384) 21 (2476.71) 23 (1351.99) -2.269
vance 0 (0) 15 (1769.08) 15 (881.731) -2.200
things 10 (1171.92) 35 (4127.85) 45 (2645.19) -2.157
state 1 (117.192) 15 (1769.08) 16 (940.513) -1.999
's 117 (13711.5) 174 (20521.3) 291 (17105.6) -1.993

In other speeches and interviews from the two of them, Vance's greater predilection for "of" vs. Walz's "'s" is consistent — about which more later — as is Walz's more frequent use of "it" and "there". Not really political but maybe of some linguistic interest.

October 2, 2024 @ 12:02 pm · Filed by Mark Liberman under Language and politics

Permalink