Questions and Answers in Parliamentary Discussions: Form and

Functions

Mare Koit

Institute of Computer Science, University of Tartu, Narva mnt 18, Tartu, Estonia

Keywords: Parliamentary Corpus, Dialogue Act, Annotation, Knowledge Representation.

Abstract: The study is aimed to develop the Estonian parliamentary corpus. The existing morphologically analyzed

corpus includes verbatim records of sessions held in the Parliament of Estonia in 1995-2001. An important

task of the Parliament is the passing of acts and resolutions. Every reading of a bill starts with a speech of a

minister and/or of a member of the responsible leading committee. Then members of the Parliament can ask

questions which will be answered by the presenter. The paper concentrates on the questions and answers that

have been annotated in the corpus according to a custom-made dialogue act annotation scheme as well as the

ISO standard. For comparison, questions and answers when reading a bill in the UK Parliament are considered.

Different forms of questions and answers with different functions are prevailing in both parliaments. The

main function of questions in the Parliament of Estonia is to get information. On the contrary, in the UK

Parliament the questions mainly are used to present arguments for or against the bill. The main function of

answers is to provide information in the Parliament of Estonia but agreement or disagreement with arguments

in the UK Parliament. Our further aim is the automatic analysis of Estonian political texts and comparison

with political discourse in other parliaments.

1 INTRODUCTION

Parliament data is a useful and practical source of

material for linguistic, political, sociological,

historical etc. research with its influential language

and content for the social and political domain. At

present, parliamentary sessions in many countries of

the world are well documented with transcripts, audio

and video recordings available online. The CLARIN

ERIC infrastructure offers access to 26 parliamentary

corpora in different languages (Parliamentary

corpora, 2021). Creating, curating and maintaining

political corpora is becoming an ever more involved

task. Such corpora must be easy to browse and search

for linguists, social scientists, digital humanists and

the general public.

In the first part of the paper, we examine

discussions on legislation in the Parliament of Estonia

– Riigikogu. Our study is based on verbatim records

of the sittings. A morphologically annotated corpus is

formed that includes the records from 1995 to 2001

(in total, 13 million tokens), both for download and

on-line searching (Koondkorpus: Riigikogu, 2021).

https://orcid.org/0000-0002-7318-087X

Since 2020, the Riigikogu is using a new system to

prepare verbatim reports of its sittings. The new

solution is based on speech recognition technology

worked out in Tallinn University of Technology. The

new system records the debates as sound files, which

the speech recognition software then transcribes.

After this, human editors revise the text, which is

finally published on the Riigikogu website

(Riigikogu, 2021).

In a previous study (Koit, 2021) we were looking

for arguments presented by the members of the

Parliament (MPs) in negotiations when proceeding a

bill in the Riigikogu. The current paper investigates

the discussions taking place before negotiation –

questions asked by MPs about the bill and the answers

of the government representatives. We are looking for

form and functions of questions and answers, using a

part of the parliamentary corpus where dialogue acts

(DAs) are annotated. We are annotating both DAs and

arguments in our corpus with the aim to make it

available the automatic analysis of political discourse

in Estonian. In the second part of the paper, as a case

study, we compare our parliamentary discussions

with the discussions in the UK Parliament.

186

Koit, M.

Questions and Answers in Parliamentary Discussions: Form and Functions.

DOI: 10.5220/0010665800003064

In Proceedings of the 13th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2021) - Volume 2: KEOD, pages 186-193

ISBN: 978-989-758-533-3; ISSN: 2184-3228

The remainder of the paper is structured as

follows. Section 2 describes related work. In Section

3, we examine a randomly chosen discussion in the

Riigikogu by using the verbatim records of three

sittings. We consider questions asked and answered

in discussions. In Section 4, by using a selected

record, we consider questions and answers in British

House of Commons when proceeding a bill. In

Section 5, we discuss the similarities and differences

between form and functions of asked and answered

questions in both parliaments. Our further aim is to

make it possible to automatically analyze Estonian

parliamentary discourse and characterize the MPs

participating in discussions by their questions,

answers and presented arguments as well as the

comparison with other parliaments. Section 6 draws

conclusions and figures out future work.

2 RELATED WORK

There are many initiatives on creation and annotation

of parliamentary corpora. Different tools are available

for the corpora that are used for various

investigations.

Eide (2020) introduces the Swedish parliamentary

debates annotated with linguistic information and

augmented with semantic links, in order to make the

data easier to use and process – in particular for

language technology research, but also for political

science and other ﬁelds with an interest in

parliamentary data.

Steingrímsson et al. (2020) introduce the

Icelandic Corpus of Parliamentary Proceedings. The

corpus has been automatically part-of-speech tagged

and lemmatized. It is annotated with extensive

metadata about the speeches, speakers and political

parties, including speech topic, whether the speaker is

in the government coalition or opposition, age and

gender of speaker at the time of delivery, references

to sound and video recordings and more.

Coole et al. (2020) compile a linguistically

annotated and semantically tagged version of the

Hansard corpus from 1803 up to the present day.

They describe the toolchain for tokenization, part-of-

speech tagging and semantic annotations.

Kerkvliet et al. (2020) train a state of the art name

identity tagger (spaCy) to recognize political actors in

Dutch parliamentary proceedings. Besides proper

nouns of persons and political parties, the tagger can

recognize quite complex deﬁnite descriptions

referring to cabinet ministers, ministries, and

parliamentary committees.

Navarretta and Hansen (2020) consider

differences in the word use of Danish parties, and

investigate how these differences can be used to

automatically identify the party of politicians from

their speeches. The analysis shows that the party of

the politicians can be distinguished in nearly 60% of

the cases, even if they debate about the same subjects

and often use the same terminology.

Hofmann et al. (2020) present a case study

comparing the lexical similarities and differences

between parties within and across two corpora of

Austrian German – a diachronic media corpus and a

corpus of parliamentary records. The results show

that changes observed in these measures can be

related to political events during that time.

Voloshchuk and Usyk (2019) study the lingo-

pragmatic features – semantic organization,

composition, and stylistic register of political

speeches. Each political speech has its specific

communicative goal, intention, and audience. The

persuasiveness as the lingual pragmatic category in

the political speeches has been analyzed.

Verbalization creates a communicative portrait of a

speaker and his own individual style.

Petukhova et al. (2015) study plenary sessions in

the UK Youth Parliament and apply the information

state update machinery to tracking and understanding

the argumentative behaviour of participants in a

parliamentary debate in order to predict its outcome.

A parliamentary debate is a communication process

in which participants argue for or against a motion.

First, segmentation has been performed together with

dialogue act annotations into functional segments

according to guidelines provided in ISO standard

(2012). To each segment a communicative function

has been assigned in one or more of the nine ISO

dimensions. An artificial agent could play different

roles in a debate, e.g. the role of one of the debaters

or their seconders by supporting or attacking certain

arguments. In the study, the agent plays the role of

concluder, whose task is to understand the arguments

of all the debaters and to conclude the debate by

stating the opinion of the majority. Its performance is

compared with a human concluder.

Chojnicka (2013) examines the use and functions

of questions in Latvian and Polish parliamentary

debates from the perspective of comparative

pragmatics. The research is based on a corpus of 200

utterances taken from transcripts of Latvian and

Polish parliaments’ sittings. It uses the typology of

questions in interaction developed by Ilie (1999) –

reaction requests, expository questions, token

information questions, suggestion questions,

evaluative/accusatory questions, and rhetorical

Questions and Answers in Parliamentary Discussions: Form and Functions

187

questions. These question types form a continuum in

terms of two features: assertiveness and interactivity.

Reaction requests are the least assertive (they do not

contain a thesis and do demand response) and the

most interactive (the response must come from

another speaker). The differences in the frequency

and functions of questions reflect different degrees of

interactivity of a debate. The discourse of Polish

parliamentary debates is more interactional than

Latvian. Another conclusion is that Latvian

parliament allows for more acute criticism and

judgement than Polish.

Bara et al. (2007) compare two approaches, one

semi-automated (Hamlet) and the other fully

automated (Alceste), when analysing debate from the

UK House of Commons on a private member’s bill

on abortion in 1966. The authors conclude that both

techniques have produced results pertinent to the

study of deliberation set within a parliamentary

context and that each of them has particular strengths.

3 QUESTIONS AND ANSWERS IN

THE PARLIAMENT OF

ESTONIA

In this section, we examine discussions in the

Riigikogu based on verbatim records of the sittings.

We consider the questions of the MPs and the answers

provided by the representatives of the government.

3.1 Empirical Material

Our empirical material is formed by the verbatim

records of the Parliament of Estonia – Riigikogu. The

records (in Estonian) are accessible on the Web (cf.

Riigikogu, 2021). An important task of the Riigikogu

is the passing of acts and resolutions. Acts are the

result of work in multiple stages. The first stage of

legislation involves the drafting of a bill. During the

second stage, the bill is initiated in the Riigikogu. The

Riigikogu conducts proceedings on bills at three

readings. The proceeding of a bill is managed by the

relevant leading committee – one of the eleven

standing committees. Passed acts are proclaimed by

the President of the Republic.

The readings have a predetermined structure.

First, the representatives of the government and the

leading committee make their presentations about the

bill and/or its amendments. After every presentation,

MPs can ask questions which will be answered by the

presenter. Then negotiation follows where arguments

for and against the bill and its amendments are given.

The 2

and the 3

readings in addition include voting

on amendments and final voting, respectively.

For this paper, dialogue acts, incl. questions and

answers, are annotated in a part of the records

belonging to the corpus that includes records from

1995-2001.

3.2 Dialogue Act Typology

An ISO standard is established for annotating DAs in

texts (Bunt et al., 2020; ISO, 2012). However, we are

also (together with the standard) using a custom-made

typology worked out for annotating Estonian spoken

dialogues (Hennoste et al., 2008). The typology is

influenced by the conversation analysis (CA).

According to CA, some DAs form adjacency pairs

(APs) where producing the first pair part makes the

second one relevant (e.g. a question requires an

answer). There are two general principles in CA

(Hutchby and Wooffitt, 1998):

1) DAs that form APs (e.g. question and answer)

must be distinguished from non-AP acts (e.g.

feedback)

2) DAs are divided into two groups: information

acts and conversation managing acts. The last

group can be divided into

a) fluent conversation managing acts, and

b) acts for solving communication problems

(repair acts).

This study concentrates on questions and answers

(AP acts). In our typology, questions are determined

as the utterances that have a specific form in Estonian:

questioning words, a specific word order and/or

intonation. There are the following types of questions

in the typology:

- questions that expect giving information – wh-

question and open yes/no question (both types

can be annotated as setQuestion in ISO

standard)

- questions that expect agreement/refusal – closed

yes/no question (propositionalQuestion in ISO)

and question that offers answer (checkQuestion

in ISO)

- questions that expect the choice of an alternative

– alternative question (choiceQuestion in ISO).

Open and closed yes/no questions have similar

form in Estonian but they expect different reactions

from the partner. A closed yes/no question expects the

answer yes or no while open yes/no question expects

giving more information, e.g. by asking the question

Is there a bus that departs after 8? customer intends

to know the departure times of buses. Both closed

yes/no question and question offering answer are

questions that expect yes/no answers. Their

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

188

difference lies in the presuppositions of the speaker.

Asking a question that offers answer the speaker has

an explicit opinion, hypothesis, and (s)he is expecting

confirmation by the partner. No such presupposition

exists in the case of a closed yes/no question.

In addition, there is also a non-AP question in our

typology – rhetorical question – as well as there are

non-AP information acts (e.g. explication,

specification, etc.). Rhetorical questions occur in

longer reports of the government representatives and

they do not expect any answer from another person.

However, we do not consider non-AP acts in the

present study.

Table 1: Question and answer DAs in two typologies

CA-

ased (ou

) ISO standar

First pair parts of APs

QUF: closed Y/N question

QUF: open Y/N question

QYF: alternative question

QUF: wh-question

QUF: offering answer

QUF: othe

Information seeking

functions

propositionalQuestion

setQuestion

choiceQuestion

setQuestion

checkQuestion

Second pair parts of APs

QUS: yes

QUS: no

QUS: agreeing no

QUS: other Y/N answer

QUS: alternative: one

QUS: alternative: both

QUS: alternative: third

choice

QUS: alternative: negative

QUS: alternative: other

QUS: giving information

QUS: missing information

QUS: refusal

QUS: postponement

QUS: alternate

QUS: othe

Information providing

functions

confirm

agreement

disconfirm

disagreement

agreement

answer

correction

answer

answe

Comparison of the two typologies – on one hand,

questions and answers in our typology and on the

other hand, information seeking and information

providing functions in ISO – is given in Table 1.

Custom-made software (Aller et al., 2014) is

being used for semi-automatic annotation of DAs in

the verbatim records of the Riigikogu. For every

utterance, the computer proposes up to five DA tags

by using 10-fold cross-validation and then an expert

disambiguates the annotation.

In our typology, names of dialogue acts consist of two

parts separated by a colon: 1) the first two letters give

abbreviation of the name of an act-group, e.g. QU –

questions, AI – additional information acts. The third

3.3 Questions Asked and Answered in

the Riigikogu

As an example, let us consider proceedings of the bill

on sale and consumption of alcohol (in 2001). The

transcripts of three sittings consist of 27,768 tokens.

In total, eight reports of two ministers (economy and

finances) and representatives of the leading

committee (economic affairs) have been presented.

After every report, MPs can ask questions.

A chairperson gives the floor, e.g. by saying Mr.

N, you have the floor. When asking a question an MP

always starts with a ritual, e.g. Thank you, Mr. Chair.

<Question to the presenter>.

In the following, we give some examples of

question-answer APs. Different first and second pair

parts of APs are presented in discussions – Example

1 QUF: wh-question and QUS: giving information;

Example 2 QUF: open yes-no question and QUS:

giving information; Example 3 QUF: closed yes-no

question and QUS: no.

ISO tags are added in the examples as well. In ISO

standard, there are two matches both to our QUS: yes

and QUS: no (Table 1). When annotating, the expert

differentiated the answers based on the first pair part

of the AP – respectively, confirm or disconfirm if it

was a closed yes/no question (propositionalQuestion

in ISO standard) and agreement or disagreement if it

was a question offering answer (checkQuestion).

(1)

Q(uestion of MP): Milliseid muudatusi sellest

eelnõust tuleneb eestimaiste jookide valmistajate,

näiteks veini- ja õlletootjate õigustes ja kohustustes?

QUF: wh-question | setQuestion

Which changes does this bill involve in rights and

obligations of producers of Estonian drinks, e.g. wine

and beer producers?

A(nswer of the Presenter): Minu arvates ei kehtesta

uus alkoholiseadus mingeid täiendavaid piiranguid

kohalikele alkoholitootjatele. <…> QUS: giving

information | answer

In my opinion, the new law does not establish any

additional restrictions for local producers. <…>

(2)

Q: <…> Öelge, palun, kas selle seaduse alusel oleks ka

politseil võimalik efektiivsemalt tegelda just niisuguse

alkoholi levitamise, hoidmise, joomisega vahelejäänud

või vahelejäävate inimeste karistamisega? QUF: open

Y/N question | setQuestion

letter is used only for AP acts – the first (F) or the second

(S) pair part of an AP act; 2) full name of the act, e.g.

QUF: open Y/N question, QUS: giving information, AI:

specification.

Questions and Answers in Parliamentary Discussions: Form and Functions

189

Please tell us, according this law, can the police act

more effectively when punishing people who

distribute, keep or drink alcohol?

A: See seadus kehtestab rangemad sanktsioonid kui

varem. <…> QUS: giving information | answer

This law establishes stronger sanctions than before.

(3)

Q: <…> Kas majanduskomisjonis oli juttu, kui palju

hakkab olema neid järelevalvet pidavaid ametnikke?

QUF: closed Y/N question | propositionalQuestion

Did the committee of economic affairs discuss the

number of officials needed for inspection?

A: <…> Konkreetsetest numbritest ei olnud juttu ja

vaevalt me saame kindlaks määrata, missugune on

optimaalne arv. QUS: no | disconfirm

We did not discuss the numbers, the optimal number

can hardly be determined.

The total number of questions asked by 95 MPs is

123. The most frequent question type is wh-question

(51 questions, or 41.5%). That is not surprising

because this question type expects giving information

(Example 1). Open yes/no question (38 questions, or

30.9%) is the other question type which expects giving

information (Example 2). As said before, an open

yes/no question has similar form as a closed yes/no

question in Estonian. When disambiguating annota-

tions, the expert takes into account how the addressee

reacts. If (s)he answers only yes or no then the question

is annotated as a closed yes/no question. If (s)he gives

more information, it is annotated as an open yes/no

question. The number of closed yes/no questions

(Example 3) is 10 (8.1%). The numbers of both

alternative questions and questions offering answer are

equal – 12 (9.8%). No questions are annotated as QUF:

other. Summing up, more than 70% of questions are

asked by MPs in order to get information. About 30%

of questions expect a short answer – yes or no or

choosing an alternative (Figure 1).

Figure 1: Types of questions (number) in the Riigikogu

when proceeding the bill on alcohol.

In the analysed records, the number of the second

pair parts of question-answer APs is a little bit less

than the number of the first pair parts – 120 vs. 123

(Figure 2). It is because two different questions

sometimes get one common answer. As expected, the

most frequent answer is QUS: giving information –

69 (57.5%). However, some of the wh-questions and

open yes/no questions will not be answered by giving

information (which is the expected reaction), but the

reaction is QUS: refusal (1 case), QUS: missing

information (7 cases), or QUS: postponement (15

cases).

Figure 2: Types of responses (number) in the Riigikogu

when proceeding the bill on alcohol.

In some cases, the missed information will later

be given by another presenter in his report, e.g. by a

member of the leading committee if the minister

redirected the answer.

4 QUESTIONS AND ANSWERS IN

THE UK PARLIAMENT

In this section, we consider a discussion in the UK

Parliament in order to compare it with discussions in

the Riigikogu. We selected the Tobacco Advertising

and Promotion Bill (2001) discussed in the British

House of Commons. We analyse the second reading

(the verbatim record includes 55,327 tokens) which is

the first opportunity for MPs for debate on the bill.

The debate is opened by the Secretary of State for

Health. The official opposition spokesperson

responds with their views on the bill. The debate

continues with other opposition parties and

backbench MPs giving their opinions. At the end of

the debate, the Commons decides whether the Bill

12 12

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

190

should be given its second reading by voting,

meaning it can proceed to the next stage.

In total, 24 presentations both of the members of

the government coalition and the opposition are

listened in the 2

reading. After every presentation,

MPs ask questions which will be answered by the

presenter. It is similar with the discussion in the

Riigikogu. We concentrate here on questions and

answers that have been annotated in the verbatim

record by an expert. Two tag sets are used for

annotation that also demonstrate the different types of

questions and answers in our CA-based typology.

In the following, there are some examples of

different types of asked questions together with given

answers – Example 4 checkQuestion and

disagreement, Example 5 propositionalQuestion and

confirm, Example 6 setQuestion and answer.

(4)

Q: Does the right hon. Gentleman not recognise that

the Government's mismanagement of these issues has

led to a huge increase in the supply of smuggled

tobacco? checkQuestion | QUF: offering answer

A: That is not the case. disagreement | QUS: no

(5)

Q: Will the Bill ban brand stretching – by which I

mean the use of brand names on items such as

clothing and other goods for sale?

propositionalQuestion | QUF: closed Y/N question

A: Yes, it will do that, but there is an important

caveat. confirm | QUS: yes

(6)

Q: If a voluntary code is as unworkable as the

Secretary of State says, why did it deliver a big

reduction in consumption between 1971 and 1996?

setQuestion | QUF: wh-question

A: There are various reasons for the reduction in

consumption. That is a long-running trend – not only

in this country but in other developed nations. <…>

answer | QUS: giving information

The number of questions asked by 61 persons is

73. The number of responses of different types is 122

(Figures 3 and 4). When agreeing or disagreeing with

the proposal set up by a check question the answerer

always gives also additional information. The most

frequent question type (i.e. information seeking

function) is checkQuestion (QUF: offering answer in

our typology) – 51 (69.9% of all questions). Out of

the remaining types, 17 (23.3%) belong to the type

setQuestion, three (4.1%) to propositionQuestion and

two (2.7%) to choiceQuestion.

The most frequent information providing function

is ‘answer’ – 75 (62.5%). After that, disagreement

(25, or 20.5%), agreement (18, or 14.8%), confirm (2,

or 1.6%), and disconfirm (1, or 0.8%) come. It should

be mentioned that five of the ‘answers’ are in our

typology annotated as ‘QUS: alternative: one’ or

‘QUS: refusal’ (three and two cases, respectively). In

addition, 63 rhetorical questions occur in

presentations (non-AP acts in our typology), but we

do not consider them here.

Figure 3: Types of questions (number) in British House of

Commons when proceeding the bill on tobacco.

Figure 4: Types of responses (number) in British House of

Commons when proceeding the bill on tobacco.

5 DISCUSSION

A comparative study of political argumentation in

different parliaments as well as in different political

cultures and different languages is a challenging

research question.

We are analysing questions asked and answered in

two parliaments when discussing a bill – the Estonian

Riigikogu and the UK Parliament House of Commons.

The legislation procedures of the parliaments are quite

different. In the Riigikogu, the representatives of

government and the leading committee make reports

about the bill. Every report will be followed by

questions of MPs. No arguments for or against the bill

are usually presented during asking and answering

questions. Debates take place in a special part of

reading – negotiation (s. Subsection 3.1). Differently,

Questions and Answers in Parliamentary Discussions: Form and Functions

191

discussions in the UK Parliament are more

complicated because the Parliament has two Houses.

We consider only one sitting in the House of Commons

– the second reading of the bill that provides an

opportunity to MPs to debate. Here, when asking and

answering questions, also arguments for and against

the bill are presented. We compare the different types

of questions and answers in two parliaments.

When comparing form and functions of questions

and answers of the two analysed proceedings, we can

see that different types of questions (and related

answers) prevail in both parliaments. In the

Riigikogu, the questions are mostly asked for getting

information (setQuestion in ISO standard) while

question that offers answer (checkQuestion) prevails

in the UK Parliament. Here, such questions are often

formulated using a specific style (e.g. does he not …),

i.e. including negation. Differently from the

Riigikogu, 3

person is used instead of 2

person,

typical expressions are e.g. Does the Secretary of

State think…, The hon. Gentleman is wrong. Check

questions actually turn out to be arguments for or

against the bill (depending on the parties of debaters).

It is different in the Riigikogu – questions and

answers do not include arguments. The arguments are

presented during negotiations where no questions are

asked (excluding rhetorical questions that do not

expect answer and procedural questions asked by the

chair of the sitting).

Therefore, the main function of questions in the

Riigikogu is to get information. The main function of

responses is to provide information. In the House of

Commons, the main function of questions is to give

arguments for or against the bill. The main function

of responses is to agree or disagree with the

arguments and to provide additional information in

order to justify the claims of arguments. The

distributions of question and answer types (in

percentage) in two compared discussions are

presented in Figures 5 and 6, respectively.

Figure 5: Types of questions (%) in two parliaments when

discussing a bill.

Figure 6: Types of responses (%) in two parliaments when

discussing a bill.

6 CONCLUSIONS

Verbatim records of sittings of many parliaments can

be accessed online. In this paper, discussions on the

bill of alcohol in the Parliament of Estonia are

analyzed in order to illustrate the types and functions

of questions and answers. For comparison, the 2

reading of the Tobacco Advertising and Promotion

Bill in UK Parliament House of Commons is

considered. A custom-made typology, based on the

conversation analysis, as well as the ISO standard is

used for annotation of dialogue acts. The main types

of questions in the Riigikogu are wh-question and

open yes-no question (both are annotated as

setQuestion in ISO) which both expect getting

information. The main type of responses is giving

information (answer in ISO standard). The main type

of questions in the UK Parliament House of

Commons is checkQuestion (a question that offers an

answer). Such questions turn out to be arguments.

The main response type is agreement or disagreement

(followed by providing additional information).

When questioning and answering in the House of

Commons, arguments for and against the bill are

given. Therefore, every report (out of 24) initiates a

debate where one participant always is the previous

presenter. That is not the case in the Riigikogu. Here

every report (out of 8) initiates an information

dialogue where similarly, one participant is the

previous presenter. Arguments are presented in a

special part of discussion – negotiation.

This study is a step towards automatic analysis of

Estonian political discussions. The current task is the

development of the parliamentary corpus where

dialogue acts are annotated. Future work includes the

ﬁnalization of the annotation process of the dataset, in

British Estonian

100

British Estonian

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

192

order to check the usability of other dialogue acts

(directives, opinions, etc.) in recognition of political

arguments, and the deﬁnition of suitable NLP

methods based on the annotated corpus. This also

makes it possible automatically to compare the

discussions in the Riigikogu with the political

discourse in other parliaments.

ACKNOWLEDGEMENTS

This work was supported by the European Union

through the European Regional Development Fund

(Centre of Excellence in Estonian Studies).

REFERENCES

Aller, S., Gerassimenko, O, Hennoste, T., Kasterpalu, R.,

Koit, M., Mihkels, K., Laanesoo, K., Rääbis, A. 2014.

Dialoogide pragmaatilise analüüsi tarkvara. In Estonian

Papers in Applied Linguistics, 23−36.

https://doi.org/10.5128/ERYa.1736-2563 .

Bara, J., Weale, A., Bicquelet, A. 2007. Analysing

Parliamentary Debate with Computer Assistance. In

Swiss Political Science Review 13(4), 577–605.

Blaette, A., Gehlhar, S., Leonhardt, C. 2020. The

Europeanization of Parliamentary Debates on

Migration in Austria, France, Germany, and the

Netherlands. In Proc. of ParlaCLARIN II, 66–74.

Bunt, H., Petukhova, V., Gilmartin, E., Pelachaud, C.,

Fang, A.C., Keizer, S., Prévot, L. The ISO Standard for

Dialogue Act Annotation, Second Edition. 2020. In

Proc. of LREC, 549–558.

Chojnicka, J. 2013. Questions in Latvian and Polish

Parliamentary Debates: A Comparative Study. In

Lingua Posnaniensis, vol. LV (1), 37–54.

https://doi.org/10.2478/linpo-2013-0003

Coole, M., Rayson, P., Mariani, J. Unﬁnished Business:

Construction and Maintenance of a Semantically

Tagged Historical Parliamentary Corpus, UK Hansard

from 1803 to the present day. 2020. In Proc. of

ParlaCLARIN II, 23–27.

Eide, S.R. Anföranden: Annotated and Augmented

Parliamentary Debates from Sweden. 2020. In Proc. of

ParlaCLARIN II, 5–10.

Hennoste, T., Gerassimenko, O., Kasterpalu, R., Koit, M.,

Rääbis, A., Strandson, K., 2008. From Human

Communication to Intelligent User Interfaces: Corpora

of Spoken Estonian. In Proc. of LREC, 2025–2032.

Hofmann, K., Marakasova, A., Baumann, A., Neidhardt, J.,

Wissik, T. Comparing Lexical Usage in Political

Discourse across Diachronic Corpora. 2020. In Proc. of

ParlaCLARIN II, 58–65.

Hutchby, I., Wooffitt, R. 1998. Conversation Analysis.

Principles, Practices and Applications. Polity Press.

Ilie, C. 1999. Question-response Argumentation in Talk

Shows. In Journal of Pragmatics 31, 975–999.

ISO: Language resource management – Semantic

annotation framework – Part 2: Dialogue acts. ISO

24617-2. Geneva, ISO Central Secretariat. 2012.

Kerkvliet, L., Kamps, J., Marx, M. 2020. Who mentions

whom? Recognizing political actors in proceedings.

2020. In Proc. of ParlaCLARIN II, 35–39.

Koit, M. 2021. How Are the Members of a Parliament

Arguing? Analysis of an Argument Corpus. In Proc. of

ICAART-2021. Vol. 2, 1046−1053.

Koondkorpus: Riigikogu, 2021. https://keeleressursid.ee-

/et/216-koondkorpus-riigikogu

Navarretta, C., Hansen, D.H. Identifying Parties in

Manifestos and Parliament Speeches. 2020. In Proc. of

ParlaCLARIN II, 51–57.

Parliamentary corpora, 2021. https://www.clarin.eu/-

resource-families/parliamentary-corpora

Petukhova, V., Malchanau, A., Bunt, H. 2015. Modelling

argumentative behaviour in parliamentary debates: data

collection, analysis and test case. In Principles and

Practice of Multi-Agent Systems. LNCS, vol 9935.

Springer, Cham. https://doi.org/10.1007/978-3-319-

46218-9_3

Riigikogu, 2021. https://www.riigikogu.-ee/en/

Tobacco Advertising and Promotion Bill. 2001.

http://hansard.millbanksystems.com/commons/2001/ja

n/22/

Voloshchuk, I., Usyk, G., 2019. Argumentation in Political

Discourse: Semantic, Composition and Stylistic

231.

Questions and Answers in Parliamentary Discussions: Form and Functions

193