Nine months of #twitterstorians

Wordcloud of the #twitterstorians corpus

Wordcloud of the #twitterstorians corpus

Last April, I started to collect tweets containing the hashtag #twitterstorians1. The idea is to question whether it is possible to capture / understand in almost real time historiographies in the making. As you all know, this hashtag was created by Katrina Gulliver.

After three months, I attempted to do a first text analysis of this #twitterstorians corpus – but it was still too tiny to be analysed. A few days ago, I attempted again to analyse it and it seemed to work, this time.

I am using a software, iramuteq, developped by Pierre Ratinaud, assistant professor in Toulouse (South-Eastern France). Iramuteq is an interface to the statistical software R. Iramuteq allow to analyse raw texts, that are stored within a corpus that is very slightly encoded (FR). By the way, iramuteq is developped in several languages (including english) and can perform text analysis in French, English, German and more. So download it and try it!

Corpus description

The corpus was divided into 24613 texts and text units (between 13 and 14 words in each text units in average – in this case, as it is a corpus made of tweets, texts and text units are the same), with 33582 occurences and 19870 forms (the corpus is lemmatised). There are 2.39% of HAPAX (forms appearing only once) and they form around 40% of the occurrences.

The analyses



The six classes

1 2 3 4 5 6
write conference amrev oupacademic military irishhistory
new call cw150 prof society onthisday
what journal civilwar associate smh_historians irelandhistory
post paper worldhistory pageant announcement birth
blog annual lizcovart bursary milhist shkenna
archive global allthingsliberty markfreeman1974 academic born
usih programme klinecj ahrc membership dublin
know online boston1775 glasgow airminded irish
great submit gettysburg fund official die
think case todayinhistory close militaryhistory invincibles
teach due morgan ahrcpress history 1916rising
love workshop edmund tudor facebook macneill
like program paul uniofnottingham hwar poet
research study revolution jan digest ireland
book proposal understand ra kansas patrick
need update revolutionary studentship jun death
find symposium earlyamhist cfp apr valera
good cfp lincoln skill 81st limerick
talk release revwar economichistory ddln found
look imperial act succession westpoint_usma leader
podcast georgian gwbooks lecturer check execution
life carnival myhnn grow mtg plunkett
work oct boston deptford mo mp
student frus egypt tmrw page eoin
acwri september liverpool kcl canberra brenmalone
social host confederate reveal date james
back pleasure apostlepaul oso phdchat parnell
will paris maier emmalautman keep hunger
survey historycarnival suffragette smoke sem casement
read follow revreborn historylabplus link cork
thought deadline pauline recordkeeping ann thomas
tip europe loyalist angelabartie list execute
robgreeneii nomination animate form opportunity belfast
past visit abraham submission crown kilmainham
source collection economics reconstruct resource series
library application landscape wcomehs sep rise
public reminder megankatenelson phd war league
question relevant myth imcl unsw stewart
hashtag report nation suit fellowship archbishop
term miss ferguson fresher prefer hang
ff oxford womenshistory age late rossa
website registration spirit britac_news net gaol
thing join civil empire job treaty
thesis the_oah gettysburg150 londoners jul wynns
man international address ryrie brian centenary
name shafrconference paint alec rape unite
19thc sexuality french durham_uni nov pearse
mean sept force project interest ruth
highered dance soldier university city roscommon

Dendogram of the six classes

Classification analysis: Dendogram of the six classes

Conclusions: academic communication and Public History

The #twitterstorians hashtag is firstly a communication channel, as the classification analysis is showing (classes 1 & 2: books, articles, blogs for class 1 and CFPs, conferences for class 2) and also the similitude one. But looking at the four other classes, you remark that there are some domains that are more dynamic: Irish history, Military history, US history. In the details, notably in the similitude graph, you observe that those tweets are first concerning Public History (see the “onthisday” linked to “irishhistory” for instance). The similitude graph shows also some hashtags that are worth a harvest: #envhist, for instance.

  1. I also collect #digitalhumanities, #digitalhistory and #dighist []

OpenEdition vous propose de citer ce billet de la manière suivante :
Frédéric Clavert (24 décembre 2013). Nine months of #twitterstorians. L'histoire contemporaine à l'ère numérique. Consulté le 21 juillet 2024 à l’adresse

5 réflexions sur « Nine months of #twitterstorians »

  1. Ping : Nine months of #twitterstorians | Fréd&e...

  2. Ping : The benefits of social media in research | Joanne Bailey Muses on History

  3. Ping : Traitement et analyse d’un corpus de tweets: sortir de l’artisanat | Frédéric Clavert

  4. Martin Grandjean

    Super intéressant ! Mais c’est vrai que le corpus est très particulier, et finalement pas si fourni que ça (il ne nous reste qu’à populariser encore le #hashtag !). Est-ce qu’il n’y a pas de contenus dans d’autres langues que l’anglais (qui est la seule à ressortir dans le tableau des occurrences, si je ne m’abuse)?

    1. Frédéric Clavert Auteur de l’article

      De temps en temps en explorant le graph de similitude, on peut voir apparaître un peu de Français ou d’Allemand. C’est très négligeable. Le corpus n’est pas si fourni que ça, mais c’est l’API de twitter: aucune garantie qu’il soit intégral ni, si c’est un échantillon, qu’il soit représentatif. C’est l’un des gros problèmes de Twitter.


Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.