ABSTRACT
This paper analyses party competition during secession crises in advanced democracies. It focuses on Quebec, Scotland and Catalonia to show how a party or parties use secession as a competition issue, how other parties respond and what are the consequences for regional party systems. Using quantitative text analysis of parliamentary debates, this work finds that during secession crises, most parties are forced to talk about territorial integrity and that party systems polarize in the territorial dimension. However, parties can adopt different strategies to alter their salience and position on the issue. Most surprisingly, this paper shows that secessionists are not always the ones that emphasize secession most and that parties do not always position at the poles of the territorial dimension.
Acknowledgements
The author is grateful to the two anonymous reviewers that offered comments on this article.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 The Discours d’Ouverture in Quebec, the Government Programme in Scotland and the Debat Sobre l’Orientació Política General in Catalonia.
2 If such debates do not take place in one year, debates for the election of the head of the regional government are used. If none existing, censorship motions are chosen. These debates follow similar procedures, they present a government programme.
3 Only debates in which all represented parties participate are collected. Speeches that belong to groups that did not originate in elections, as splits, are not collected. If two government representatives intervene, the second speech is considered a party speech. If two speakers of the same party share intervention time, they are considered one speech.
4 Due to the different working languages in each parliament, three dictionaries in different languages have been created. Almost all speeches in the Scottish Parliament are in English, and in the Quebec National Assembly in French. Most speeches in the Catalan Parliament are in Catalan. Speeches in Spanish are translated with Google Translate. This should not be a problem since the focus is on word frequencies and not on phrase meaning. For an example, see (Leonisio and Strijbis Citation2012) and (De Vries et al. Citation2018).
5 A priori, terms that are likely to be expressed by parties for and against secession (e.g. secession, break-up of the state) are included in all dictionaries deductively.
6 More words that belong to the general concept of secession in each particular territory are added to each dictionary inductively.
7 This is a percentage resulting from the number of words related to the dictionary used in the speech of a party divided by the total number of words in the speech, multiplied by 100.
8 In order to identify them, all debates that can relate to these issues are selected from the headlines of the minutes of sessions of each regional parliament.
9 For instance, these would be the Scottish National Party and the Scottish Conservative and Unionist Party in Scotland. Each party gets the values 1 and −1. When selecting parties, it is better that they are present during several years to make comparisons easier.
10 LBG rescaling is applied with smooth = 1. Stop-words are not deleted and terms are not stemmed because there is no theoretical justification for it. Otherwise, some terms that are crucial for scaling could be lost.
11 Dalton calculates polarization with voters’ average location of parties in a left-right scale that ranges from 0–10 and then applies a weighted mean using parties’ vote shares. Here, the scaled positions of party speeches and these parties’ parliamentary seat shares are used instead. Thus, this measure of polarization must not be interpreted as voter polarization, but as the polarization of the party system: it varies as a combination of the number of parties, their positions and their weights. If the government participates in a debate, its position is weighted by the number of seats that elected that government, and polarization is calculated dividing by the total seat-share of the parliament plus those seats. Regional electoral results come from Schakel’s Regional Election database (Citation2013), which has been completed with data provided by regional parliaments.
12 Readers are welcome to contact the author to access the dictionaries and data.
13 No territorial debate was found for 1975.
14 Centre here means an equidistant position to the two extremes of the territorial dimension.
15 There are no entries for 1963–1966, 1999, 2001, 2006, 2011, 2012, 2014 and 2015 because not all parties in parliament participated in debates.
16 There is no entry for 2005.
17 There are no entries for 1984, 1988 and 1991.