Hella Grits - In Maps
December 15, 2016 8:05 PM   Subscribe

The great American word mapper
Where the top 100,000 words are used the most, as seen through Twitter data
The data for these maps are drawn from billions of tweets collected by geographer Diansheng Guo in 2014. Jack Grieve, a forensic linguist at Aston University in the United Kingdom, along with Andrea Nini of the University of Manchester, identified the top 100,000 words used in these tweets and how often they are used in every county in the continental United States, based on location data from Twitter.
posted by hilaryjade (45 comments total) 12 users marked this as a favorite
 
I proudly live in the tri-state euchre bubble.
posted by Thorzdad at 8:19 PM on December 15, 2016 [12 favorites]


I was trying to find terms exclusive to the Gulf coast, so "Fondren" worked better than expected...
posted by eustatic at 8:23 PM on December 15, 2016


Bus vs Freeway has kind of an amusing Northeast-Southwest thing to it
posted by BinGregory at 8:29 PM on December 15, 2016


Highly recommend turning OFF regional smoothing. It's inventing data and begging the question-- assumes a certain spatial homogeneity in language and uses the result of that to qualitatively support that assumption.

GIS interpolation is a really hard problem, and can only be solved correctly by AT LEAST taking the demographics of counties into account.
posted by supercres at 8:31 PM on December 15, 2016 [8 favorites]


Detroit is the premier dude-bro city, who knew?
posted by BinGregory at 8:31 PM on December 15, 2016




Really, though, the smoothed maps are just not believable if you've looked at a lot of these-- linguistic trends by county. The non smoothed versions are really quite amazing.

Now if only we could do something about that whole population/area mismatch (so for analyses like these, available data/area mismatch). If only we could get people to read cartograms.
posted by supercres at 8:36 PM on December 15, 2016 [1 favorite]


St. Louis and Milwaukee are really midwest outliers for 'soda'
posted by T.D. Strange at 8:38 PM on December 15, 2016 [1 favorite]


The biggest overlap between "sushi" and "biscuit" appears to be in northeastern Montana...
posted by thefoxgod at 8:43 PM on December 15, 2016


"Happy" vs "sad" has a strong east-west component.
posted by plastic_animals at 8:47 PM on December 15, 2016


Northeastern Montana has sushi, biscuits and lots of swearing. If it weren't so fucking cold, I would consider moving there.
posted by ArbitraryAndCapricious at 8:51 PM on December 15, 2016 [1 favorite]


The biggest overlap between "sushi" and "biscuit" appears to be in northeastern Montana...

Also true for 'porno' and 'bible'
posted by T.D. Strange at 8:52 PM on December 15, 2016 [3 favorites]


Spend some time with pray, prayer, Jesus, God, church, religion. Pretty interesting variations.
posted by Miko at 9:02 PM on December 15, 2016


sad vs. happy
posted by idiopath at 9:17 PM on December 15, 2016


I'm trying to figure out what's going on in northeastern Missouri and the Yasss vs. Yolo Convergence
posted by drlith at 9:23 PM on December 15, 2016 [1 favorite]


the groupings for BUTTS are fascinating
posted by poffin boffin at 9:28 PM on December 15, 2016


I would love to dig into the data a bit. Even though it was gathered in 2014, I'm sure there's something to be learned by running a script to determine the two words that map most closely to the 2016 presidential election results.
posted by turbowombat at 9:45 PM on December 15, 2016 [2 favorites]


Hmmm......
posted by Navelgazer at 10:15 PM on December 15, 2016 [1 favorite]


Soooo I'm guessing that CHUD means something else around the lakeshore of Ohio and Indiana and not that they have a problem with cannibalistic humanoid underground dwellers?
posted by ROU_Xenophobe at 10:31 PM on December 15, 2016 [1 favorite]


As a person who was born in the South but spent the majority of my adult life on the West coast, I have probably said "hella grits" before.
posted by scose at 1:06 AM on December 16, 2016 [5 favorites]


So am I getting this right in thinking the maps reflect how often a word is used on Twitter by people in X county compared to other counties without reference to population?

So, for example, the NYC map without smoothing shows a strange little hotbed of activity up around Plentywood Montana, where, one might assume, there could be a couple people tweeting at each other about taking a trip to NYC at a frequency higher than mentions in the rest of the populations of other counties due to greater volume of subjects from counties other than Plentywoods?

If so, it sounds like low population areas would get a lot greater and more unusual variance in high volume word usage than cities where there are a lot more tweets drowning out total usage of a word by greater numbers of words used overall.
posted by gusottertrout at 2:06 AM on December 16, 2016 [1 favorite]


There is no data for "Fuckety." I didn't think I existed anyway.
posted by Mr. Yuck at 2:40 AM on December 16, 2016 [1 favorite]


Happy to see that while "brunch" is predominantly a northeast/southwest thing and "grits" purely a southwest thing, my town is a hotbed of both.

Come visit Durham, NC! We'll have grits with brunch!
posted by ardgedee at 4:03 AM on December 16, 2016


(^of course grits is a southeast thing, not a southwest thing. oops.)
posted by ardgedee at 4:26 AM on December 16, 2016


META FILTER
posted by Rock Steady at 4:31 AM on December 16, 2016


I was hoping to be responsible for a central Ohio "Wicked" bump, but alas, I must try wicked harder.

This would be fun in conjunction with that NY Times lexicon quiz.
posted by ChuraChura at 4:40 AM on December 16, 2016 [1 favorite]


Are they're just a ton of Pittsburgh transplants in Casper, WY or...?
posted by soren_lorensen at 4:41 AM on December 16, 2016


There is no data for "Fuckety." I didn't think I existed anyway.

That's more of a Saskatchewan thing in my experience ...
posted by oheso at 5:31 AM on December 16, 2016



I was trying to find terms exclusive to the Gulf coast, so "Fondren" worked better than expected.


"Pawpaw" did ok. "Hurricane" produced a bit odd results.
posted by RolandOfEld at 5:42 AM on December 16, 2016


There's a Hurricane Ridge out on the Olympic peninsula. And the PNW windstorms we get in the winter are often hurricane force winds, so I'd imagine that accounts for the dark spot near Seattle for "hurricane."
posted by los pantalones del muerte at 5:55 AM on December 16, 2016 [1 favorite]


How do I get this thing to differentiate between "miːfaɪ" and all the other, wrong ways that you people say it.
posted by FirstMateKate at 6:12 AM on December 16, 2016 [1 favorite]


The distribution for "BOOTY" is truly remarkable. Given this and the "BUTTS" data noted above, just what exactly is going on there in Montana?
posted by JimInLoganSquare at 7:44 AM on December 16, 2016


The state most infatuated with tweeting about Krishna? Utah. Go figure.
posted by kozad at 7:48 AM on December 16, 2016


Grits and brunch broke my brain. Brits and grunts britches and lunch brats and grunts.
posted by idiopath at 8:06 AM on December 16, 2016


I'm guessing some of the Montana hotspots occur because the calculations take into account the surrounding counties, which are (even more) sparsely populated. Hill County, MT has a population of ~16500, whereas the next county west (Liberty) has only ~2400. Hill County has an actual city (Havre), where people are congregated, and Liberty County don't have nothin' except farms. About a quarter of Blaine County, on the east side of Hill County, is made up of a reservation.

So a few people in Havre tweeting about (e.g.) "Coke" can really throw off the region. ("Booty" seems to be centered in Fergus County, another "populous, dense" region by northern MT standards.)

If you've spent your life on the coasts, you really should go to the interior, it is so incredibly isolated you won't believe it.
posted by AFABulous at 8:57 AM on December 16, 2016 [2 favorites]


(Here are all the Indian reservations!)
posted by AFABulous at 8:59 AM on December 16, 2016 [1 favorite]


HEINEKEN? FUCK THAT! PABST. BLUE. RIBBON!
posted by cmoj at 9:13 AM on December 16, 2016


Southerners damn. Northeasterners goddamn. And apparently there's one pissed off dude in the top right corner of Nevada.
posted by telepanda at 9:54 AM on December 16, 2016


What's so funny, Arizona?
posted by AFABulous at 10:00 AM on December 16, 2016


"Haha" vs "LOL" is surprisingly northwest/southeast
posted by saturday_morning at 10:07 AM on December 16, 2016


The "regional smoothing" option is making my inner statistician cringe hard. It makes absolutely no sense, and it's damaging the data. As mentioned above, yes, turn off this feature.
posted by late afternoon dreaming hotel at 10:07 AM on December 16, 2016


What is going on with Dead? Do people not die west of the Mississippi?
posted by Dojie at 12:28 PM on December 16, 2016


Metafilter: Inexplicable booty hotspots
posted by Sebmojo at 2:48 PM on December 16, 2016


Lutefisk — no data. Hm.
posted by traveler_ at 9:40 PM on December 16, 2016


There is no data for "Fuckety." I didn't think I existed anyway.
posted by Mr. Yuck at 2:40 AM on December 16 [+] [!]


Fuckity
posted by mefireader at 10:15 AM on December 19, 2016


« Older What happened to the gold from the S.S. Central...   |   “Otus is presented as an agent, not as a victim or... Newer »


This thread has been archived and is closed to new comments