It's like a warzone out there\": Tweeting place in the Tottenham riots of 2011

“It’s like a war zone out there”: discursive constructions of place in Tweets during the Tottenham riots of 2011

Dr B. Walker Dr I. R. Lamond

Background to dataset

Inherited Big Data.

In search of a narrative place and considering how this changes during a period of civil unrest

Two researchers – learning from each other and sharing approaches

A iterative, layered coding approach

A computational corpus linguistics approach

Initial research question: How is a narrative of place constructed in the data set

2011 1985

An iterative, layered coding approach (Qualitatively led)

• Refinement of timeframe (from first Tweet overtly linked to Tottenham riots to 12 hours later)

• Listing Tweets alphabetically to highlight Retweet (thereby identifying unique messages)

• Reading through those Tweets several times to gain an insight into developing a qualitative list of

broad themes by which to initially code them.

• Drawing out themes for specific attention

• (for this presentation) Merging lists - removing Retweets and placing back into chronological order.

• Reading through those Tweets to establish a more refined coding.

• Drawing on key moments reported by the mainstream news media (Guardian and BBC).

• Tabulating that coding and graphically representing that as a means of highlighting shifts in

narrative.

A iterative, layered coding approach

A iterative, layered coding approach

A iterative, layered coding approach

A iterative, layered coding approach

To 22

To 23

To 00

To 01

To 02

To 03

To 04

To 05

To 06

To 07

T1 in

T2 in

> T2

T3 in

T4 in

T5 in

> T5

>> T

5>>

> T5

T6 in

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

D M B H W S > <

T1 = Police car first reported as on fire; T2 = Bus first reported as on fire; T3 = Fire engines reported as coming into riot area to control fires; T4 = News teams under attack from rioters; T5 = Reports of significant increase in looting; T6 = Reports of situation under police control

D = Tottenham is a Dump; M = The people of Tottenham/ The authorities are just Mad; B = It’s just a place where Bad things happen and Bad people live; H = Tweet exhibits Humour; W = Contains a description or allusion to violence to a person, authority or property; S = Suggests Sympathy for those rioting or level of understand for the situation of the area in which the rioting is taking place; > = References a historical context; < = Ties to suggest or place the riot in a wider global context.

A iterative, layered coding approach

• M B H < present throughout

• Strong showing of H at a number of points

• Greater instance of comments pertaining to M and B, over D

• Most occurrences of W tie in with periods when news media recorded specific higher profile

Instances

• Lower percentage showing for S until later in the night’s events

• Greater diversity of sub-themes earlier in the sample (earlier on in the nights events)

NB: No single narrative of place

A computational corpus linguistics approach (Quantitatively led)

Analysing data subsets• Wordlists generated using AntConc (Anthony, 2014)

• Wordlists = all the words (types) in the file listed in order of frequency (or alphabetical order).

Wordlist• 6th, 7th and 8th August 2011• 279370 tokens in the twitter data• 19171 types

A computational corpus linguistics approach

Refining Wordlist for type selection:

– Most frequent type = Tottenham (obviously)– Many other high-frequency types features of the data

• http, co, com, t (http://t.co/xHcdzAp)• rt, bit, ly, twitpic

– Others due to one or two Tweets reTweeted• may, looter, posing, regret, mark, duggan, shooting

– Mark Duggan may not have been using gun before he was shot dead– Tottenham Looter may later regret posing with his loot– Tottenham riot over Mark Duggan shooting

– Pronouns• It, I• Possibly interesting for future investigation of language of twitter, but not

directly relevant to this study.

A computational corpus linguistics approach

• People (no. 54 in the list, 699 occurrences)• First type of potential interest (after dismissing first 53 on grounds just discussed)• Words investigated using concordance lines.

• people used in a variety of Tweets in different ways

• Looking at people over 3 days:– 06/08/11 – 153 instances– 07/08/11 – 453 instances– 08/08/11 – 93 instances

People used in Tweets that offer:– Information and observation

(including news report)– Discussion and evaluation

A computational corpus linguistics approach06/08/11• Information and observations about the riots

Tottenham people are setting police cars on fireAfter cops sped three riot vans thru crowd, people now sensibly building barricadesShop front and number of police cars aflame in #tottenham many people standing and watching, police on horsesDozens of people outside police station in Tottenham have burnt bins & trying to attack police line.

• Evaluation and discussion about the riots: for / justification

People are angry in the UK. Corrupt police, corrupt government, corrupt press. #tottenham #racism #cuts #fightback#tottenham its the polices fault ! The systems Corrupt! Then they wonder why people riot!Difficult to dismiss people at these protests as 'violent thugs' - there are huge & complex problems with poverty & police racism #tottenham

• Evaluation and discussion about the riots: against / no justification

So people riot in #Tottenham because police shot a guy with a gun that shot at them? #madness #idiots

A computational corpus linguistics approach07/08/11• Information and observations about the riots

Video of people literally queuing up to loot JD Sports in #WoodgreenSome of the buildings gutted by fire in #tottenham had flats above them. They don't know if everyone got out. People have lost everything.London Ambulance Service: 10 people treated, 9 taken to the hospital with injuries related to Tottenham riot

• Evaluation and discussion about the riots: for / justification

Bullet lodged in police radio was police issue. And people wonder why the cops are hated and not trusted.Will the #Tottenham incident spread? It's the overall trend, not the flashpoint of incident: No jobs, angry people vs. gov't & police.Black people are SEVEN times more likely to be stopped by Police than White people.

• Evaluation and discussion about the riots: against / no justification

Whatever you think of the riot, some people lost their home and their livelihoods last night, and that is awful. #tottenhamno excuse for the #tottenham riots,they burned down people homes and livelihoods.if the police can't deal with them they army can #EDL #UAF

A computational corpus linguistics approach08/08/11• Information and observations about the riots

42 people arrested, 26 police officers injured in riots in Tottenham area of London#Protest Tottenham riot 2011: 50 protesters arrested, buildings smolder:nick cleggs in tottenham talking to the police, how about you//go broadwater farm or tiverton and talk to the PEOPLE? #londonriots• Evaluation and discussion about the riots: for / justification

• Evaluation and discussion about the riots: against / no justification

nick cleggs in tottenham talking to the police, how about you//go broadwater farm or tiverton and talk to the PEOPLE? #londonriotsPeople are recommending Ken Livingstone's appearance on the BBC. He's blamed the #Tottenham riot on spending cuts:

Did people see Tottenham on the news after the riot and think.. *Thats Exactly How I Want My Area To Look*Why can't they just loot 10 Downing St instead of affecting the REAL people! #Walthamstow #Tottenham #Enfield #Brixton #LondonRiotsA strong message to police and politicians: the people of Tottenham demand justice. Or xboxes and booze. Whatever. Let's burn down a shop. it astounds me that people would loot their own community, #brixton & #Tottenham are hardly paved with gold you have just f**ked yourself!

A computational corpus linguistics approach

• Top keywords (most statistically significant words that occur more on 06/07 than on 07/08 than you would expect by chance)

on, cars, fire,

police, week, station,

Thursday, man, now,

apparently, guy, killed,

called, car, pool,

outside, pigs, cop,

bottles, set, blood,

road, attacked, connected,

right, along, files,

kid, warned, ransacked.

Broad observations on the project so far

Raises several questions about genre.

Raises questions concerning the writer’s understanding of audience.

No dominant narratives are coming to the fore – however a number of clusters around narratives seem to be emerging. Though our skill set feels limited in the face of the data set….we do seem to be making some progress.

Numerous difficulties emerge when trying to conduct research on such a wide ranging set of sources for a data set…with our current skill set and relative restricted knowledge of higher end computational analysis we struggled to find appropriate software that could support us in making sense of the narratives drawn from such a wide spectrum of sources.

Need for data constructed lexicon to support a computationally driven network/cluster analysis of the corpus.

Developments of our research question:Comparing Twitter and media construal of identity

Possibilities emerging from examining the riots through a lens of critical event studies.

• ‘Event’ conceptualised as disruption

• Essential contestation in events and the

articulation of the contested construal of

the event.

• Who is encountering the event as

disruption

• How is the disruption being articulated

• Who is trying to manage the disruption

• How is the disruption being ‘managed’