This codebook contains the instructions necessary for the semi-automated annotation task in the project ‘Years of turmoil: The Political Consequences of the Financial and Economic Crisis in Europe’. The first section defines which information in the news wires we read is to be considered relevant in terms of this protest event analysis. The second section details how different references to the same protest event have to be linked. The third section lists and explains the indicators that have to be annotated with respect to every single protest event. Finally, in the fourth section, the annotation interface is described and precise coding instructions as well as examples are specified.

Table of Contents

1. Relevance of protests: Location | Time | Form of action

2. Identification of events

3. Indicators

4. Manual:

Instructions: Dates | Locations | Number of participants | Action forms | Issues | Economic aspects | Actors

Interface: Annotation | Recoding | Restoring | Report

5. Appendix: List of parties | List of trade unions


Which documents and sentences have to be annotated?

The decision on whether a document or sentence is relevant for this project depends on whether a relevant protest event is covered in the document. We define a relevant protest event on the basis of three types of information: location, time and action form.


A protest event is only relevant for us if it takes place in one of the following countries we are interested in (all EU countries minus Croatia plus Switzerland, Norway and Iceland):

Austria, Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, United Kingdom

In most cases, there is an explicit link between the reference of a protest and a location, e.g. “Thousands demonstrated yesterday in Madrid…” or “The Helsinki protests led to…”, sometimes, however, the location can be inferred from persons or organizations.


A protest event is only relevant for us if it happened in the years between

January 1, 2005, and December 31, 2014

Mostly, a reference to a date (such as the year, month or even day) or date-like information (such as “yesterday”, “last week” etc.) is explicitly linked to a protest event. In the cases in which such information is not present throughout the whole document, the date of publication of the article has to be assumed to be the date of the protest.


Our conceptual goal is to collect data on all politically motivated, unconventional actions in the selected countries and time period. In practice, we do not rely on a precise theoretical definition of a relevant protest action but on a detailed list of unconventional or non-institutionalized political action forms:

Action form Additional explanation
Strike industrial action of any kind (incl. work stoppages, pickets)
Collections of signatures and petitions incl. a collectively signed letter or collective letter with is send individually
Demonstrations and marches protesters take to the streets, they move
Protest camps/meetings/ vigils/rallies and other festive forms people do not move, the protest is staged in one place
Symbolic protest actions e.g. street theatre, performances
Boycotts incl. consumer boycotts, school boycotts
Cyber-attacks incl. mail bombings, hacking
Hunger strikes
Refusals of payment loans, taxes, etc.
Blockades incl. sit-ins and picket lines
Squattings Occupation of land or houses
Bomb threats
Symbolic violence against objects or persons e.g. paint bombs, tomatoes, eggs, burning books
Other confrontational actions e.g. mutiny of prisoners, shouting at the president or at any authoritative figure, threats against persons, whistle-blowing
Destruction of private or public buildings breaking windows, break-ins, burglary, etc.
Bomb or arson attacks
Violence against persons incl. kidnapping, persons injured or killed
Clashes with police

For the assessment of the relevance of a document or sentence, it is sufficient that a reference to a protest event corresponds to one of the actions listed.


A rule of thumb is that political protest is ‘bottom up’. Actions which are conducted as part of usual institutional processes are thus not considered in this analysis. This refers, for example, to police operations, court cases or all kinds of legislative procedures. Also, protests by political authorities of whatever kind do not count as protest events.


With the exception of bomb threats, we exclude threats or planned events that are announced without any specified date. If the date is specified, we include them.


We don’t evaluate whether a protest is legal or not. Every political protest, which meets an entry in the list of action forms above, is relevant.


How do we identify protest events?

A relevant document may contain references to one protest event only, or it may contain references to more than one protest event. Moreover, the same protest event may be referred to several times at different places in the document. The events in the document can be distinguished on the basis of the following ‘form-time-locality’ rules:

  1. Protests of different action forms define different events, e.g. the combination of a general strike with a demonstration in the capital city on the same day would define two separate protest events.

  2. Protests of a given action form define different events, if they take place at different times, i.e. different days or clearly distinct moments on the same day (e.g. a series of demonstrations by the same actor, but on successive days) and/or in different locations (e.g. parallel demonstrations for the same goal that take place on the same day, but in different cities, correspond to as many protest events as cities are mentioned in the document). Different events of the same form (e.g. strike) taking place on the same day at different places in the same city count as one event, even if they are mentioned separately (e.g. strikes by different actors, such as metro drivers and airport controllers).


To identify individual protest events, we apply the same list of action forms that we also use for the evaluation of the relevance (see action forms).

A particular problem is posed by what we call “campaign references”, which are general, summarizing references to several protest events. An example is the following sentence: “Today, teachers in three different parts of Spain protested.” According to our rules to separate protest events, we would expect to annotate several events, i.e. one for each of the “three different parts of Spain”, but there is no explicit information on these locations. There are two alternatives how we have to deal with this problem:

  1. If the story presents explicit information on all the single events, e.g. “in Madrid, 10’000 teachers gathered in front of the parliament, while about 5000 took to the streets in Valencia and Barcelona”“, we annotate the three separate events, one each in Madrid, Barcelona and Valencia.

  2. If it is clear that the ‘campaign reference’ does mention much more than the references to the single events, or if there is no further reference specifying the single events at all, we take this campaign reference as a separate protest event. In the example, we would annotate a protest event in Spain on the basis of the ‘campaign reference’ if less than three events are more detailled described.


Enumerations of protest events, e.g. ‘in 80 cities people took to the streets’ or this is the ‘5th bomb attack this year’, are considered as single protest events, if no further information on locations and dates is given.


An event which does qualify as a protest event on the basis of the form-time-locality rules, is not coded as a protest event. If the actor performing the event is an ‘elite’-actor, who performs the action
- as part of his daily routine (e.g. an interest association protesting (in a press conference, by writing a letter or by lobbying a decision-maker)
- exceptionally, but without mobilizing non-elite individuals (e.g. French mayors protesting against the law on gay marriages and refusing to implement it), or
- in inter-state relations (e.g. cyber-attacks by Russia against the Baltic states).


You do not have to annotate previously annotated events. However, if a new reference to a previously annotated event contains additional information, you should add this new information to the annotation of this event.


The following table shows the list of indicators that have to be annotated per protest event as well as their descriptions.

Label Description
Event index Index indicating separate events
Date of event Date of event as specified in the document.
Location of event Geographical reference associated with the protest event.
N. of participants Information on the size of protest event in terms of participants.
Action form Action form applied during a protest event.
Goal of protest Political goals associated with the protest event.
Economic aspects Whether there is an expicit economic aspect to an issue (‘New social movement’ and ‘Regionalism’ only).
Actors Political groups organizing or participating in a protest event.


The event index refers to the number of the event as identified in step 2 (identification of events).




Most of the events should be quite straightforward to annotate. However, there are certainly also difficult decisions during the annotations, for which the following guidelines might be helpful.

  1. Overinterpretation: We try to objectivize the information we read in the text as well as possible. This means that we only annotate what is explicitly mentioned and abstain from adding information from our common knowledge or interpretations of the information. For example, if we know that thousands of persons take part in a general strike (because, obviously, a general strike involves the entire economy of a country), but we do not explicitly read about this in the story, we do not enter this information on the number of participants. In a similar vein, if it is written in a story that a demonstration in a city is ‘similar’ in terms of the number of participants than another demonstration in another city, we do not enter this information if there is no other, explicit description of the number of participants. However, please, note that we use our own knowledge about detailed locations in order to code the name of cities (e.g. if ‘Acropolis’ is mentioned, but no city name, we code ‘Athens’ for the city, if we know that the Acropolis is located in Athens; or if ‘Cotroceri Palace’ is mentioned and we know that this is in Bukarest, we code ‘Bukarest’).

  2. Missing information: To define a protest event, only an action form, a location (which can also be an entire country) and a time (which can also be an extended time period) need to be explicitly mentioned. All other indicators are ‘optional’ in the sense that we can leave them empty if no information is provided for them.

  3. Ambivalent information: If there is ambivalent information, we rely on the description of protests by the journalist. For example, if a story reports about ethnic violence, but a police report that is mentioned suggests that it is a criminal act (so no relevant protest), it depends on whether other information in the story explicitly links the violence to ethnic conflicts or not. If yes, the event nevertheless is a relevant protest.

During the annotation, each document has to be read twice:

In the first reading, all references to relevant protest events have to be identified. There are pre-annotated sentences, which are highlighted in yellow were found on the basis of relevant language patterns by an automated recognition, but you also have to search for sentences that are not pre-annotated but may still be relevant. Moreover, words describing locations are highlighted in green, words referring to times are highlighted in blue, and names of persons and organizations are highlighted in red.


The first reading also serves the purpose to identify the different references, which refer to the same protest event. So please think about which references belong to the same event while doing this first reading.


Every event automatically receives an index, i.e. a number that allows us to separate the different events. During the annotation, you have to enter all variables at the level of events. Hence, you have to consider the information from all references on the same event for the coding of the indicators.

In the example shown in the following table, the first and second references do not refer to the same event, since there is a clear difference in time between the two events mentioned (“last year”). These references have different event indices (values 1 and 2). The references three and four, however, belong to the same event, which is indicated with the same event index of value 3.

References to a protest event Event index
“On Monday, several thousand students protested against the planned raise in tuition fees.” 1
“It’s the largest student protests since the demonstrations last year.” 2
“Truckers blocked the main entrance to the city.” 3
“50 trucks brought traffic on the highway to a standstill.” 3

During the second reading, the labels of the six indicators date, location, number of participants, action form, issue and actors have to be annotated for every protest event.


If a document contains no relevant protest, you have to click the irrelevant story checkbox before going to the next story.


If possible, we always indicate the full date in the format yyyy-mm-dd, i.e. the day, month and year separated by a point. This means that also indirect information on the exact date (e.g. “yesterday”) have to be translated into an actual date with help of the date of publication. Hence, for the example of “yesterday”, if the date of publication is 02-09-2006, the date entry is 01-09-2006.


If the date is not clear fom the text, we enter the date of publication of the news story.


If only a longer time period is indicated (e.g. “the sit-in from last Wednesday on”, “the bombings in August 2003” or “the riots in 2003”), there are three ways to enter the date, depending on the information provided:
1. If a starting day is apparent as in the first example, we take this as the date of the protest event.
2. If no starting day is given, but the time period is equal or less than a week, we take the Monday of the corresponding week as date of the protest.
2. If no starting day is evident and the time period is longer than a week, we have to enter the time as is into the entry field named ‘Non-standard date’, e.g. ‘2003-08’ or ‘2003’ for the second and third example above. If possible, convert the dates you enter in the ‘Non-standard date’ field as number with the same format as the date in the date entry field.


We enter all different geographic names related to a protest event into the first text field. If possible, please indicate only the name of the city where the protest takes place. E.g. if a story reports on a ‘bomb attack at Belfast airport’, enter ‘Belfast’ and not ‘Belfast airport’ into the location entry field. In the second list selection field, you additionally have to select the country.


Here, all information on the size of a protest needs to be entered. This information can be present in the documents as numbers (e.g. ‘50 workers’) or words (e.g. ‘one thousand protesters’). Also, the information may be precise or only vague (e.g. ‘several demonstrators’ or ‘thousands of teachers’). Please use the following rules to enter the number of participants:

  1. If the number is precise, please enter it in digits. Thus, ‘thousand’ would have to be entered as ‘1’000’.

  2. If the information is vague, please use the conversion indicated in the following table.

example of reference number to enter
several, some, a few, a group of, a couple of 5
dozens, a number of 50
hundreds 500
thousands 5’000
tens of thousands 50’000
hundreds of thousands 500’000
millions 5’000’000


Only persons which are actively involved in the protests are considered participants. Thus, neither the police, spectators nor victims are considered as participants.


If there is more than one piece of information given, take the one that is more precise (e.g. if ‘hundreds’ and ‘600’ are mentioned in relation to the same protest event, enter ‘600’). If there are multiple pieces of information at the same level of precision (e.g. one number is published by the police and a different number by the organizers), please take the average.


We enter unclear indications such as ‘more than 3000’ or ‘over 20’000’ defensively. The examples thus are entered as 3’000 and 20’000, respectively. If there are indications on the size of protests you cannot locate in the conversion table such as ‘many’ or ‘a large crowd’, please leave the entry field ‘number of protesters’ empty and copy-paste the correpsonding string plus a remark into the comment field.


One of the following labels has to be selected for the action form indicator.

Label Description
Strikes industrial action of any kind (incl. work stoppages, pickets)
Demonstrations Demonstrations, marches, rallies, meetings, vigils and other non-confrontational gatherings
Petitions and related activities petitions, letters, consumer activism, boycotts, symbolic protests (performances etc.)
Blockades and related activities blockades, occupations, sit-ins, camps and other confrontational strategies that are related to a specific place
Violent protests Sabotage, riots, destruction of private or public buildings, bomb or arson attacks, violence against persons, clashes with police, cyber-attacks
Other protests


For the annotation of the issues, one of the following labels has to be selected:

Label Description
Economics (private) Economic claims addressed to firms/ employers: dismissal of staff, closure of firm/branch, labor conflict related to pay rise, pay cut etc.
Economics (public) Economic claims addressed to public institutions, e.g. welfare, budget policies, agricultural policies, labor regulation, state regulation of the economy more generally
Environment Environmental protection, Nuclear energy, Other forms of energy production, Infrastructure projects (e.g., transport), Animal rights
Cultural liberalism Peace (questions of war & peace, nuclear and other conventional weapons, military infrastructure, military spending, military service), Women’s rights (incl. equal treatment, abortion), LGTB (rights and recognition of lesbians, gay, transsexuals, bisexuals), International solidarity (development aid; anti-imperialism), Anti-racism, rights of migrants more generally, Squatters mobilization (for autonomous living and cultural spaces)
Regionalism Separatism, regional independence, protection of regional interests, anti-regionalist counter-mobilization
Cultural conservatism counter-mobilisation to all aspects of new social movement except for anti-racism and migrants’ rights (which is in xenophobia)
Xenophobia Right-wing extremism, racist mobilization (against foreigners or ethnic minorities), anti-Immigration
Political Representation, corruption, electoral reform, institutional issues in general
Others All other issues not covered by the previous categories


There can be none, one or multiple issues per event. In case of multiple issues, you should select all categories that apply.


For the non-economic issue categories, the specific claims can sometimes have an economic aspect as well. Since, in this research, we are most interested in the question, how strong protests on the economy are, we want to separately gather this information. An example for a ‘New social movement’ issue, which has a distinct economic aspect, is wage inequality between women and men. As for ‘Regionalism’, for example issues related to taxes or fiscal equalization between regions and the central government have a distinct economic aspect. In cases where there is an explicit reference to economic aspects, you therefore have to check the radio button ‘economic aspects’.


We code all actors that are reported to call for, take part in or organize a given protest event. If you find such an actor, one of the following labels has to be selected:

Label Description
Parties (left) Political parties from the left (e.g. Social Democrats, Greens or Communists)
Parties (right) Political parties from the right (e.g. Liberals, Christian Democrats, Conservatives, Nationalists or Agrarians)
Parties (unknown) Political parties for which the ideological orientation is not clear
Unions (private) Private sectors unions
Unions (public) Public sector unions
Unions (both) Unions for both the private and public sector
Unions (unknown) Unions for which sector affiliation is not clear
Other Organizations Residual category for all other types of organized actors such as NGO’s
Social groups (occupational) Workers, teachers, lawyers, journalists, nurses
Social groups (students)
Social groups (pensioners)

The categories contain two different types of actors. Categories 1 to 7 cover formal organizations, whereas the other categories cover (unorganized) social groups. In the case of organizations, we distinguish three types: political parties, trade unions, and all other types of organizations. Please refer to the lists in the Appendix for an overview over parties and unions.


There can be none, one or multiple actors participating in an event. In case of multiple actors, you should select all categories that apply.

Political parties are divided into left-wing and right-wing parties. In case of doubt about this classification, use category ‘political party (unknown)’. The decision to code parties as left-wing or right-wing can be based on explicit references in the text or the list of political parties provided in the appendix, see list of parties.


Note that the references in the text might explicitly refer to the political orientation of the parties (left-wing or right-wing) or to party families. We categorize party families as follows: 1. Left-wing = Social Democrats, Greens, New Left, Communists, Radical left 2. Right-wing = Conservatives, Christian Democrats, Liberals, Radical right

Unions are divided into public, private and both sectors unions. In case of doubt, use category ‘Unions (unknown)’. Again, the decision to code unions as public or private can be based on explicit references in the text or on the list of unions in the appendix, see list of unions.


Note that references in the text might usually refer to the occupational groups organized by the union(s). Public sector covers all groups employed by the state, such as civil servants, police officers, teachers, doctors, nurses or fire fighters. Private sector covers all occupational groups employed by private businesses, such as industrial workers, service sector employees, miners, taxi drivers, etc.

Social groups are not organized groups of persons, which can be distinguished by a relevant societal characteristic. In these annotations, we only focus on the labor market position occupation, i.e. groups of people which are characterized by a job or by unemployment (=‘Social groups (occupational)’), as well as on pensioners (=‘Social groups (pensioners)’) and students (=‘Social groups (students)’).


Single persons and very general descriptions (e.g. ‘people’, ‘persons’ or ‘protesters’) are not considered as social groups.


If no specific organization is mentioned, we annotate actors as social groups. Hence, ‘doctors’ is a social group. If an ‘Association of doctors’ is mentioned, we annotate this actor as an organization.


The annotation is done in an online interface called “YoT-A”. It can be reached at You can work with any browser.


If you are annotating from outside the UZH network, you need to establish a VPN tunnnel first.

On the webpage, you’ll see a list of links, one of which should be your last name. Click on it to login. Upon login, you’ll see the entry page shown below.

YoT-A entry page

There are five panels:
1. Annotation: Annotate protest events here.
2. Recoding: Make corrections on and get an overview over your previous annotations. 3. Restoring: Restore annotated stories and get an overview over the stories to annotate. 3. Report: Fill out your weekly report.
4. Instructions: This file.


In the annotation panel, you have the entry fields for the single indicators on the left and the news story (incl. document ID, news wire and date of publication) on the right (see image below). When starting an annotation session, you have to get a first report by clicking ‘next story’. Then, you should enter the information on all indicators for the first event on the right. If you are done, click ‘save event’ to save your annotation on this event and open a new form for a potential further event. If there is no further event, click on ‘next story’ to finish this document and start with a new one.