One of the major research interests for political science has always been the study of political discourse and parliamentary debates. This literature review offers an overview of the most prominent research methods used in political science when studying political discourse. We identify the commonalities and the differences of the political science and corpus-driven approaches and show how parliamentary corpora and corpus-based approaches could be successfully integrated in political science research.

As part of the project ParlaMint II, a new corpus of the sessions of the Portuguese Parliament from 2015 to 2022 has been compiled, encoded and annotated following the ParlaMint guidelines. We report on the contents of the corpus and on the specific nature of the political settings in Portugal during the time period covered. Two subcorpora were designed that would enable comparisons of the political speeches between pre and post covid-19 pandemic. We discuss the pipeline applied to download the original texts, ensure their preprocessing and encoding in XML, and the final step of annotation. This new resource covers a period of changes in the political system in Portugal and will be an important source of data for political and social studies. Finally, Finally, we have explored the political stance on immigration in the ParlaMint-PT corpus.

This paper employs the ParlaMint-ES-GA dataset to scrutinize the intersection of gender, speech, and representation within the Parliament of Galicia, an autonomous region located in North-western Spain. The research questions center around the dynamics of women’s participation in parliamentary proceedings. Contrary to numerical parity, we explore whether increased female presence in the parliament correlates with equitable access to the floor. Analyzing parliamentary proceedings from 2015 to 2022, our quantitative study investigates the relationship between the legislative body’s composition, the number of speeches by Members of Parliament (MPs), and references made by MPs in their speeches. The findings reveal nuances in gender representation and participation, challenging assumptions about proportional access to parliamentary discourse.

The paper discusses some fine-tuned models for the tasks of part-of-speech tagging and named entity recognition. The fine-tuning was performed on the basis of an existing BERT pre-trained model and two newly pre-trained BERT models for Bulgarian that are cross-tested on the domain of the Bulgarian part of the ParlaMint corpora as a new domain. In addition, a comparison has been made between the performance of the new fine-tuned BERT models and the available results from the Stanza-based model which the Bulgarian part of the ParlaMint corpora has been annotated with. The observations show the weaknesses in each model as well as the common challenges.

Recent work in political science has made exten- sive use of NLP methods to produce evidential sup- port for a variety of analyses, for example, inferring an actor’s ideological positions from textual data or identifying the polarisation of the political discourse over the last decades. Most work has employed variations of lexical features extracted from text or has learned latent representations in a mostly un- supervised manner. While such approaches have the potential to enable political analyses at scale, they are often limited by their lack of interpretabil- ity. In the talk, I will instead look at semantic and pragmatic representations of political rhethoric and ideological framing and present several case stud- ies that showcase how linguistic annotation and the use of NLP methods can help to investigate dif- ferent framing strategies in parliamentary debates. The first part of the talk investigates populist framing strategies, specifically, the use of pronouns to create in- and out-groups and the identification of people-centric messages. The second part of the presentation focusses on framing strategies on the pragmatic level.

We present a new dataset, , that provides valuable insight for advancing discourse analysis of parliamentary debates in Portuguese. This is achieved by processing the open-access information available at the official Portuguese Parliament website and scraping the information from the debate minutes’ PDFs contained therein. Our dataset includes interventions from 547 different deputies of all major Portuguese parties, from 736 legislative initiatives spanning five legislatures from 2005 to 2021. We present a statistical analysis of the dataset compared to other publicly available Portuguese parliamentary debate corpora. Finally, we provide baseline performance analysis for voting behaviour classification.

The paper describes the process of preparation of the Polish Round Table Corpus (Pol. Korpus Okrągłego Stołu), a new resource documenting negotiations taking place in 1989 between the representatives of the communist government of the People’s Republic of Poland and the Solidarity opposition. The process consisted of OCR of graphical transcripts of the talks stored in the form of parliament-like stenographic transcripts, carrying out their manual correction and making them available for search in a concordancer currently used for standard parliamentary transcripts.

In this paper, we use automatic language identification to investigate the usage of different languages in the plenary sessions of the Parliament of Finland. Finland has two national languages, Finnish and Swedish. The plenary sessions are published as transcriptions of speeches in Parliament, reflecting the language the speaker used. In addition to charting out language use, we demonstrate how language identification can be used to audit the quality of the dataset. On the one hand, we made slight improvements to our language identifier; on the other hand, we made a list of improvement suggestions for the next version of the dataset.

This paper demonstrates the research potential of a unique European Parliament dataset for register studies, contrastive linguistics, translation and interpreting studies. The dataset consists of parallel data for several European languages, including written source texts and their translations as well as spoken source texts and the transcripts of their simultaneously interpreted versions. The paper presents a cross-linguistic, corpus-based case study on a word formation phenomenon in these European Parliament data that are enriched with various linguistic annotations and metadata as well as with information-theoretic surprisal scores. It addresses the questions of how initialisms are used across languages and production modes in the English and German corpus sections of these European Parliament data, whether there is a correlation between the use of initialisms and the use of their corresponding multiword full forms in the analysed corpus sections and what insights on the informativity and possible processing difficulties of initialisms we can gain from an analysis of information-theoretic surprisal values. The results show that English written originals and German translations are the corpus sections with the highest frequencies of initialisms. The majority of cross-language transfer situations lead to fewer initialisms in the target texts than in the source texts. In the English data, there is a positive correlation between the frequency of initialisms and the frequency of the respective full forms. There is a similar correlation in the German data, apart from the interpreted data. Additionally, the results show that initialisms represent peaks of information with regard to their surprisal values within their segments. Particularly the German data show higher surprisal values of initialisms in mediated language than in non-mediated discourse types, which indicates that in German mediated discourse, initialisms tend to be used in less conventionalised textual contexts than in English.

In making official transcripts for meeting records in Parliament, some edits are made from faithful transcripts of utterances for linguistic correction and formality. Classification of these edits is provided in this paper, and quantitative analysis is conducted for Japanese and European Parliamentary meetings by comparing the faithful transcripts of audio recordings against the official meeting records. Different trends are observed between the two Parliaments due to the nature of the language used and the meeting style. Moreover, its diachronic changes in the Japanese transcripts are presented, showing a significant decrease in the edits over the past decades. It was found that a majority of edits in the Japanese Parliament (Diet) simply remove fillers and redundant words, keeping the transcripts as verbatim as possible. This property is useful for the evaluation of the automatic speech transcription system, which was developed by us and has been used in the Japanese Parliament.

In this paper, we test the efficacy of using GPT-4 to annotate a dataset that is the used to train a BERT classifier for emotion analysis. Manual data annotation is often a laborious and expensive task and emotion annotation, specifically, has proved difficult even for expert annotators. We show that using GPT-4 can produce equally good results as doing data annotation manually while saving a lot of time and money. We train a BERT classifier on our automatically annotated dataset and get results that outperform a BERT classifier that is trained on machine translated data. Our paper shows how Large Language Models can be used to work with and analyse parliamentary corpora.

We are going to describe the Open Parliament TV project and more specifically the work we have done on alignment of video recordings with text proceedings of the german Bundestag. This has allowed us to create a comprehensive and accessible platform for citizens and journalists to engage with parliamentary proceedings. Through our diligent work, we have ensured that the video recordings accurately correspond to the corresponding text, providing a seamless and synchronised experience for users. In this article, we describe the issues we were faced with and the method we used to solve it, along with the visualisations we developed to investigate and assess the content.

This article resorts to mixed methods to examine British and Spanish parliamentary discourse. The quantitative corpus-assisted (lexical priming) theory and data are complemented by the qualitative discourse historical approach. Two CLARIN ParlaMint corpora – ParlamMint-GB and ParlaMint-ES – are queried in the analysis, which focuses on English (“Rusia” and “Ukraine”) and Spanish (“Rusia” and “Ucrania”) nodes and collocations. In sum, the analysis sketches a brief profile of each corpus. The British House of Commons is more homogenous, strongly associating “Russia” and “Ukraine” with their participation in the war. Furthermore, this chamber shows a greater interest in “Russia. The Spanish Congreso de los Diputados indicates greater quantitative differences (heterogeneity). Here, “Russia” clearly transcends its role as a military contender and is also portrayed as an economic competitor for the West. Unlike in Britain, the Spanish lower house shows more mentions of “Ucrania”, which is assigned just one role – as an invasion victim. In conclusion, the productivity of corpus-assisted mixed methods is confirmed along with the precious value of the ParlaMint constellation.

We introduce a dataset on political orientation and power position identification. The dataset is derived from ParlaMint, a set of comparable corpora of transcribed parliamentary speeches from 29 national and regional parliaments. We introduce the dataset, provide the reasoning behind some of the choices during its creation, present statistics on the dataset, and, using a simple classifier, some baseline results on predicting political orientation on the left-to-right axis, and on power position identification, i.e., distinguishing between the speeches delivered by governing coalition party members from those of opposition party members.

The paper introduces the IMPAQTS corpus of Italian political discourse, a multimodal corpus of around 2.65 million tokens including 1,500 speeches uttered by 150 prominent politicians spanning from 1946 to 2023. Covering the entire history of the Italian Republic, the collection exhibits a non-homogeneous consistency that progressively increases in quantity towards the present. The corpus is balanced according to textual and socio-linguistic criteria and includes different types of speeches. The sociolinguistic features of the speakers are carefully considered to ensure representation of Republican Italian politicians. For each speaker, the corpus contains 4 parliamentary speeches, 2 rallies, 1 party assembly, and 3 statements (in person or broadcasted). Parliamentary speeches therefore constitute the largest section of the corpus (40% of the total), enabling direct comparison with other types of political speeches. The collection procedure, including details relevant to the transcription protocols, and the processing pipeline are described. The corpus has been pragmatically annotated to include information about the implicitly conveyed questionable contents, paired with their explicit paraphrasis, providing the largest Italian collection of ecologic examples of linguistic implicit strategies. The adopted ontology of linguistic implicitness and the fine-grained annotation scheme are presented in detail.

We demonstrate the multilingual search engine and Ngram viewer that was built on top of the Parlamint dataset using the recently available translations. The user interface and SERP are carefully designed for querying parliamentary proceedings and for the intended use by citizens, journalists and political scholars. Demo at https://debateabase.wooverheid.nl. Keywords: Multilingual Search, Parliamentary Proceedings, Ngram Viewer, Machine Translation

This paper has two objectives: to present (a) the creation of ParlaMint-GR, the Greek part of the ParlaMint corpora of debates in the parliaments of Europe, and (b) preliminary results on its comparison with a corpus of Greek party manifestos, aiming at the investigation of the ideologies of the Greek political parties and members of the Parliament. Additionally, a gender related comparison is explored. The creation of the ParlaMint-GR corpus is discussed, together with the solutions adopted for various challenges faced. The corpus of party manifestos, available through CLARIN:EL, serves for a comparative study with the corpus of speeches delivered by the members of the Greek Parliament, with the aim to identify the ideological positions of parties and politicians.

This paper describes the ParlaMint 4.0 parliamentary corpora as made available in TEITOK at LINDAT. The TEITOK interface makes it possible to search through the corpus, to view each session in a readable manner, and to explore the names in the corpus. The interface does not present any new data, but provides an access point to the ParlaMint corpus that is less oriented to linguistic use only, and more accessible for the general public or researchers from other fields.

Historical parliamentary debates offer a window into the past and provide valuable insights for academic research and historical analysis. This paper presents a novel web application tailored to the exploration of historical parliamentary corpora in the context of Slovenian national identity. The developed web viewer enables advanced search functions within collections of historical parliamentary records and has an intuitive and user-friendly interface. Users can enter search terms and apply filters to refine their search results. The search function allows keyword and phrase searching, including the ability to search by delegate and place names. It is also possible to search for translations of the text by selecting the desired languages. The search results are displayed with a preview of the proceedings and highlighted phrases that match the search query. To review a specific record, the full PDF document can be displayed in a separate view, allowing the user to scroll through the PDF document and search the content. In addition, the two corpora of Slovenian historical records integrated into the viewer—the Carniolan Provincial Assembly Corpus and the Parliamentary Corpus of the First Yugoslavia—are described and an insight into the corresponding preparation processes is provided.

Entity Linking is a powerful approach for linking textual data to established structured data such as survey data or adminstrative data. However, in the realm of social science, the approach is not widely adopted. We argue that this is, at least in part, due to specific setup requirements which constitute high barriers for usage and workflows which are not well integrated into analyitical scenarios commonly deployed in social science research. We introduce the dbpedia R package to make the approach more accessible. It has a focus on functionality that is easily adoptable to the needs of social scientists working with textual data, including the support of different input formats, limited setup costs and various output formats. Using a ParlaMint corpus, we show the applicability and flexibility of the approach for parliamentary debates.

The Japanese House of Representatives, one of the two houses of the Diet, has adopted an Automatic Speech Recognition (ASR) system, which directly transcribes parliamentary speech with an accuracy of 95 percent. The ASR system also provides a timestamp for every word, which enables retrieval of the video segments of the Parliamentary meetings. The video retrieval system we have developed allows one to pinpoint and play the parliamentary video clips corresponding to the meeting minutes by keyword search. In this paper, we provide its overview and suggest various ways we can utilize the system. The system is currently extended to cover meetings of local governments, which will allow us to investigate dialectal linguistic variations.

This paper provides insight into automatic parliamentary corpora development. One year ago, I created a simple set of tools designed to continuously and automatically download, process, and create corpora from speeches in the parliaments of European Union member states. Despite the existence of numerous corpora providing speeches from European Union parliaments, the tools are more focused on collecting and building such corpora with minimal human interaction. These tools have been operating continuously for over a year, gathering parliamentary data and extending corpora, which together have more than one billion words. However, the process of maintaining these tools has brought unforeseen challenges, including issues such as being blocked by some parliaments due to overloading the parliament with requests, the inability to access the most recent data of a parliament, and effectively managing interrupted connections. Additionally, potential problems that may arise in the future are provided, along with possible solutions. These include problems with data loss prevention and adaptation to changes in the sources from which speeches are downloaded.

In this paper, we address government and opposition speeches made by the Danish Parliament’s members from 2014 to 2022. We use the linguistic annotations and metadata in ParlaMint-DK, one of the ParlaMint corpora, to investigate some characteristics of the transcribed speeches made by government and opposition and test how well classifiers can identify the speeches delivered by these groups. Our analyses confirm that there are differences in the speeches made by government and opposition e.g., in the frequency of some modality expressions. In our study, we also include parties, which do not directly support or are against the government, the “other” group. The best performing classifier for identifying speeches made by parties in government, in opposition or in “other” is a transformer with a pre-trained Danish BERT model which gave an F1-score of 0.64. The same classifier obtained an F1-score of 0.77 on the binary identification of speeches made by government or opposition parties.

Detecting opinions, their holders and targets in parliamentary debates provides an interesting layer of analysis, for example, to identify frequent targets of opinions for specific topics, actors or parties. In the paper, we present GePaDe-ORL, a new dataset for German parliamentary debates where subjective expressions, their opinion holders and targets have been annotated. We describe the annotation process and report baselines for predicting those annotations in our new dataset.

This position paper makes an argument for creating a corpus similar to that of ParlaMint, not consisting of parliamentary proceedings, but of documents released under Freedom of Information Acts. Over 100 countries have such an act, and almost all European countries. Bringing these now dispersed document collections together in a uniform format into one portal will result in a valuable language resource. Besides that, our Dutch experience shows that such new larger exposure of these documents leads to efforts to improve their quality at the sources. Keywords: Freedom of Information Act, ParlaMint, Government Data