How people cited the Bible on Twitter, April 2009-February 2010. Discusses popular authors, author locations, popular verses, popular verses, cross-references, and how people used the Bible to respond to a few events during the year.
I use the Twitter APIs to grab everything that might be a Bible reference—string matching. Use Bayesian-style filtering to avoid false positives. PS3 (Playstation 3 vs. Psalm 3) is hard to parse since many titles make religious references. Output is at website.
Increases 5x: 2,000 per day in April 2009 to 11,000 per day in February 2010
Probably matches Twitter growth rate, but no data to support
@Replies: someone sends a Bible verse to someone else. @MileyCyrus is a frequent recipient (1200 times), mostly warning her to keep her life in order. Retweets are when someone posts a Bible verse and other people retweet it. Normal tweets are just Bible citations plus additional text.
The number of retweets has increased largely because celebrity pastors have joined Twitter.
Over 180,000 people only tweeted one Bible verse. One person tweeted over 50,000—we’ll talk about him later.
People on the right tend to tweet more but are retweeted less, while people in the top left tweet less but are retweeted less. The top left tends to be celebrities (Joel Osteen, Joyce Meyer), or verse-of-the-day-style applications.
81% U.S.
Religion in California tends to be a Southern California thing. Pretty much tracks population centers. The low New York City nubmers may be a data-quality issue.
New York and the Northeast are less than you’d expect based purely on pouplarity.
Only tracking English-speaking tweets, so you expect to see blips in English-speaking countries.
This is a subset.
There are a few extreme outliers.
Posts 400 verses per day.
Rick Warren is twice as efficient as John Piper at converting retweeters. Rev Run used to be in rap group Run DMC.
Piper is big in Minnesota and the southeast. Rev Run is big in large population centers. He’s bigger than Rick Warren in Southern California, which is where Warren is. Warren doesn’t have any clear patterns domestically…
But is big in Indonesia. Piper is unusually popular in Brazil—I don’t’ know why. Rev Run has followers in the Caribbean. Piper’s version of choice is dominantly the ESV; Rev Run is KJV, but less dominantly; Warren is more evenly distributed: primarily NIV, NLT, MSG.
Jer 29:11, Phil 4:13, Rom 8:28, and John 3:16 are far and away the most popular. Nothing in Genesis-Deuteronomy or the Minor Prophets.
Swaths in Psalms, Proverbs, Paul. Jeremiah 29:11 provides almost all the popularity of Jeremiah 29. Large parts of Kings, Chronicles, and the Prophets aren’t popular. If you look at one of these charts for each day and animate it, you can see the progression of popularity starting in Genesis on January 1 as people go through their reading plans.
Jittery. MSG overrepresented. KJV surge in October is the result of TD_Jakes, who happened to cite it explicitly. NLT surge in December is a result of Rev Run, who doesn’t usually cite a translation. MSG is overrepresented based on sales.
YouVersion is dominant. You can tell they added Tweet functionality to their app in October. Crossway is unusually prominent—I don’t think they explicitly encourage tweeting, so these are organic links.
Caveat: YouVersion is so dominant that this chart largely reflects YouVersion users.
ESV is overrepresented here based on market share.
The ESV is only unambiguously the ESV 12% of the time. It could also be the NASB 20% of the time.
NASB has a surge in June 2009 based on one person tweeting hundreds of NASB verses over a few days and then stopping. The NIV surge on January 1 is because of Rev Run.
It’s said that you’d be able to reconstruct all except sixteen verses from the NT solely from quotations in the church fathers. Can we do that from Twitter? Not so much—the best we can reconstruct is 27% of the NIV.
OT is blue; NT is red. There’s not a lot of differences among the translations. It follows the overall popularity.
When someone refers to two separate verses in Twitter, could we get useful cross-reference data from the data in aggregate? Compare to existing cross-reference systems to see. Genesis cross-reference to Genesis is in the bottom left. Genesis cross-reference to Revelation is in the top left. Revelation cross-reference to Genesis is in the bottom right.
Genesis halo: similar stories of Abaraham, Isaac, and Jacob. Axis: most cross-references are to verses close to each other. Trident: parallel stories in Kings and Chronicles. Prophets: many prophets lived during the Kings-Chronicles period. Proverbs: internal links in Proverbs. Gospels: paths in Matthew, Mark, Luke; John has internal links. Revelation: heavily interlinked.
Very similar.
Very similar.
Kind of similar.
A few commonalities. But only 12% (2,400) of these 20,000 overlap existing cross-reference systems. The other cross-reference systems overlap each other 50-75%, suggesting they may have a common ancestor (but no proof for this). So Twitter isn’t a good source of cross-references, but some of the interlinks are interesting: tend to be more devotional than historical or thematic.
The most popular verse related to the Haiti earthquake. JohnPiper tweeted this verse. For Chile, there were fewer tweets: teen actor Nathan Kress from iCarly had the most retweets related to Chile.
That this was a meme on Twitter made it to the press.
People tweeted this verse to show that even God was tired of the hype around New Moon.
These are the most popular words. They don’t follow the most popular searches on Bible Gateway very well. Data useful for publishers: identify translation mindshare. Useful for developers: transiently improve verse popularity (a search for New Moon around the time of the movie release should show Isaiah 1:14 as the top result). Useful for translators: Philippians 4:13 (“I can do all things through Christ who strengthens me”) is quoted 2/3 of the time in the KJV rather than other translations because it says “through Christ.” Modern translations say “through him,” following the Greek. So if you’re interested in having verses taken out of context, don’t care what the original language says, and want to cater to popular opinion, you could translate using this data.