How do cities speak about themselves on Twitter?

Marseille speaks in&out

Team Members

Ludmila Girardi, Agata Ludzis-Todorov, Marta Severo.


We understand the internet as a vast space of topological metrics where actors can relate from all over the world, where we have a social space made pertinent on a Global scale (LÉVY, 2003; 2008). The relations produce other types of spaces and they form “places of reticular metrics” (BEAUDE, 2012). So, how to identify these new kind of places? The internet rapports a societal “problem”: living within this almost unlimited connectivity, cities' spaces are now "hybrid", where territories are more dense and more capable to have the complementarity of this common global space to relate, so the terrain (the physical space) and the territories or cities’ administrative limits are no longer essential to understand society. We need to identify where a based society relate in these new kind of places with others limits and hierarquies.

The networks now have the same importance as territories on studying a place under a spatial dimension, even more in the context of urban and city studies, since they have the tradition to rank and to compare hierarchies, centralities, postulating the relevance of studying topological spaces' metrics. Because of the emergence of this new “huge scale”, Lévy (2008) understands places to be now comparable to others in the world in terms of distance - to define places proximities in other metrics than the Euclidean system -, and in terms of scale - to define the limits or regions by the phenomenon’s substance. So, in the context of the present study, we may ask which are the links that connects places and which are the new borders of a city?

Within this framework we are studying cities as a “complete society” (LÉVY, 2003), by which we tried to identify on this sample project what are the connections "within" and "without" the French city of Marseille, considering places mentions in one of the main social media on the internet, the Twitter, at a specified space-time period (See CitySpeaks group's project context). Therefore, we were interested on Marseille's network of places from a global and from on a local scale perspective.

As a huge and vast space, the internet also addresses the empirical problem of the amount of data available to download. How can we filter all that information so we can grasp the details? “Content curation” is a trend also visible evident in places and cities studies (WILKEN, 2014). In Twitter, hashtags (#) are an essential thing to organize and filter information, so it is a very important thing researchers must consider when analysing digital data. Hashtags are aggregators of content, which is a way to study the profile of individual social media users based on their physical activity.

Social media users exhibit and archive physical experiences alongside other identity markers and employ the “spatial self” as a way to communicate where they are/were, or what they are/were doing, as well as who they are (SCHWARTZ and HALEGOUA, 2014). This type of selection of check-ins demonstrates a way social media users coordinate and incorporate the places displayed in their profile with other presentations of self. The authors encourage social media researcher “to think about place and space as strategically chosen markers of identity” (SCHWARTZ and HALEGOUA, 2014).

Research Questions

How and from where does Marseille on-site users speak in Twitter?

From where users speak about Marseille in Twitter, and in what languages?

Specific questions:

1. What are the places correlated to Marseille in twitter conversations?

2. How strong are their connections, considering the frequency of mentions?

3. What are the places and languages speaking about Marseille on a global and on a local scale?

4. What are the scales of users tweeting from Marseille accordingly to their specified location?


We conducted an exploratory analysis of Twitter data using a Cartographic approach as empirical method, specifically to organize database towards mapping comparison. According to Lévy (2008), the visual resonance of maps with each other is a cartographic means that considers the unity of the world. The objective with this empirical method was to overlay city’s scales in terms of types of languages, location and places correlation of Marseille’s mentions in Twitter between a short period in the dataset collections, from 23rd to 29th of June 2014.

We have worked with two different data collection. First, we extracted all the tweets using hashtags to identify which one was the most frequent and the occurrence of places in hashtags. Second, we extracted full datasets available: the keyword database, which considers all mentions of Marseille in all European languages, and the geolocation database, which considers all tweets generated within Marseille limits.

1. Hashtags analysis (digital curation):

a. #marseillelife: top hashtag;

b. #places: Marseille’s correlation with places in different scales considering the frequency;

2. Full datasets filtering:

a. Keyword database – organized by user specified language;

b. Geolocation database – organized by user location and classified in different scales (filters sequence: exclusives registers, categories of places).


First, we analysed the #marseillelife, which is a username on Twitter also a local blog on Tumblr that curates geotagged Instagram pictures, co-created by many authors. The whole creation of the blog and hashtag is based on intermedia transfer and confront the multi-level use of links to design and multiply the content.

Places are the common point of every post and users activity in the blog and # because they are geotagged, so we can identify all the places in Marseille city and surroundings that are areas of interest and with web complementarity.

The tweets’ texts containing the #marseillelife indicate a close relationship between users and local environment. The visualized discourse shows numerous references to characteristic places in the city, like Cassis, a coast city frequented by locals and tourists.

Users’ discourses on #marseillelife mainly reproduce places of leisure that may be touristic areas. It is the good life in the city being talked about: “love”, “summer”, “sun”. Users write more in English about than in french, but they use rather similar words: “amour”, “soleil”, “vie”.

Additionally, analysing all hashtags containing name of places mentioned in the tweets, we found specific places inside the city at an inter-urban scale, including hashtags for each quartier, which the most frequents of mentions are for the more central ones.

We should also say the hashtags emphasize Marseille’s geographical location in the region of Provence (Provence, Vitrolles, Pennes-Mirabeau), addressing an indication of places’ connectivity. There are other French cities with hashtags, especially those more close to Marseille in terms of cultural relations and physical proximity.

Considering the previous text analysis, we built four maps to compare them with each other, two focusing on the local scale and other two considering the global scale, although these two can also be comparable on a local scale.

Map 1. The sense of the local (#marseillelife)

a. Mostly generated in places of circulation: airport, roads, ship, beaches (Cassis), port.

b. Cluster in the city center: relevance of the port to the city.

Map 2. Marseille world places relations: #places by frequency of mentions

a. Brazil is due to the World Cup.

b. Although there are not many mentions in the USA/Canada they seem to have a relevant connection with Marseille.

c. Physical proximity matters a lot (West and Central Europe), but also the substance of historical background considering the connection with North Africa countries, which were French colonies, especially Algeria which has more mentions to Marseille than the Capital Paris. Algeria and Marseille have also important port relations.

Map 3. World speaking about Marseille (by user specified language)

a. The use of language is attached to the territory, although there is a spread pattern of voices in English (as the global language) and in French (the local language in Marseille).

b. There is a similar distribution of the previous map, specially the same countries in the North of Africa.

c. In the local scale (zooming in the city) we have an indication of locals (considering the National scale as local) from outsiders (English, which we cannot precise from where they are, but there are some Dutch Users, maybe indicating a connection between The Netherlands and Marseille).

Map 4. Scales of users specified location

a. The tweets are not only located within Marseille’s city limits, but within Provence, which we consider as a local scale in the categorization, from a perspective of the world system of cities (on a global scale the region has a local relevance).

b. We can differentiate French users and local users from the others. We see a concentration of more outsiders than locals in the airport and touristic spaces, especially La Ciota, where there’s a predominance of world users.

c. In the city center there is, as expected, a miscellaneous of users, which gets weaker in cities’ suburbs with the predominance of local users.

d. After filtering the geo database, we were able to collect local users’ conversations identified by the fact that it does not have a specific topic but common words in French, in opposition of the world topic conversations, which were about the World Cup.


Hashtags give the sense of the local as well as other ways of filtered digital content. Marseille has hashtags for each city’s zones and people report to specific places in Marseille, not the city. That way, it is easier to identify users and conversations about local themes, more than global topics and subjects, likewise the World Cup.

These data are more appropriate for researching an event occurance. On that matter, it is mandatory to select a precise range or time period that explains the issue, so content analysis could bring more relevant findings.

The geotagged content’s production concentrates in the city centre and mobility places, leaving the affirmation of importance of location because of the availability of internet access. This concentration also affirms that availability of time matters: to produce content, one needs time to do it, so people generates tweets during trips’ spare time.

To differentiate locals from outsiders, the location and the language combined give a fair indication. It is not easy to identify user identity based on these data, so it shall be better to filter the data in the sense of using the most precise ones and working with clean samples, taking off the “noises” of big data.

The maps do not explain great things alone, but they combined brought a fair understanding about what the city of Marseille is on this Twitter, locally and globally. To understand the local one must consider the global, and to compare them outcomes a re-scaling of cities’ limits. Marseille is connects itself by Twitter to some parts of the world accordingly to different substances, redefining borders and reinforcing bounds.


BEAUDE, Boris. Internet, changer l'espace, changer la société. Les logiques contemporaines de synchorisation. France : FYP éditions, 2012.

LÉVY, Jacques. Quels espaces pour la société-monde? Le | 25.06.2003 à 18h50 • Mis à jour le 18.03.2005 à 16h39. Avaiable on []. Last access on 10/07/2014.

LÉVY, Jacques. L’invention du Monde: une Géographie de la mondialisation. Paris: Sciences Po Les Presses, 2008.

SCHWARTZ, R., HALEGOUA, G.R. The spatial self: Location-based identity performance on social media. New Media & Society, 2014. DOI: 10.1177/1461444814531364. < >

WILKEN, R. Twitter and Geographical Location. Katrin Weller, Axel Bruns, Jean Burgess, Merja Mahrt and Cornelius Puschmann (eds.), Twitter & Society, New York: Peter Lang, 2014, 155-167.
Topic revision: r5 - 18 Jul 2014, LudmilaGirardi
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback