Title of the Paper [16 point font]


© Sung-Jun Pyon & Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

A Network of Argumentation Schemes and 

Critical Questions 

SUNG-JUN PYON 

Faculty of Philosophy 

KIM IL SUNG University 
Ryomyong Street, Taesong District, Pyongyang 

The Democratic People’s Republic of Korea 

phil2@ryongnamsan.edu.kp  
 

YONG-SOK RI  

Faculty of Philosophy 

KIM IL SUNG University 
Ryomyong Street, Taesong District, Pyongyang 

The Democratic People’s Republic of Korea 

ys.ri@ryongnamsan.edu.kp  

 
Abstract: In this paper, we devise a 

network that consists of argumenta-

tion schemes and critical questions 

that participants in debates can use to 

easily construct arguments that attack 

or support former arguments. As a 

prototype, we build a potential net-

work of argumentation schemes and 

critical questions with a practical 
reasoning scheme at its center. The 

usefulness of a NASCQ in construct-

ing and reconstructing complex 

arguments and in formal argumenta-

tion is also explored along with 

argumentation more broadly. 

Résumé: Dans cet article, nous 

concevons un réseau composé de 

schémas d'argumentation et de ques-

tions critiques que les participants aux 

débats peuvent utiliser pour construire 

facilement des arguments qui at-

taquent ou soutiennent d'anciens 

arguments. En tant que prototype, 

nous construisons un réseau potentiel 
de schémas d'argumentation et de 

questions critiques avec un schéma de 

raisonnement pratique en son centre. 

Nous explorons l'utilité d'un NASCQ 

dans la construction et la reconstruc-

tion d'arguments complexes, dans 

l'argumentation formelle et également 

dans l'argumentation en général.

 
Keywords: argumentation scheme, critical question, complex argument, formal 

argumentation 

 

mailto:phil2@ryongnamsan.edu.kp
mailto:ys.ri@ryongnamsan.edu.kp


788 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

1. Introduction 

Most argumentation begins with an argument being put forward 

that defends the standpoint adopted by the proponent. Then, the 

other participant(s) concerns will be most likely twofold: to find 

any weaknesses in the former argument and to construct a subse-

quent argument that attacks or supports the former. Essentially, 

“what is a weakness of this argument?” and “how can I construct 

an argument that attacks or supports the former?” are two key 

questions asked by the proponent and the respondent. The propo-

nent tries to find answers to these questions for the sake of bolster-

ing their former argument and in turn defending their standpoint, 

while the respondent does so in order to cast doubt on the propo-

nent’s standpoint and encourage them to reject it. In order to find 

adequate answers to these questions, both the proponent and re-

spondent can use argumentation schemes and their attendant criti-

cal questions. 

Argumentation schemes are molds for arguments that can be 

used to construct real arguments. More precisely, argumentation 

schemes are forms of argument (structures of inference) that repre-

sent structures of common types of arguments used in everyday 

language discourse, as well as in special contexts like those of 

legal argumentation and scientific argumentation (Walton et al. 

2008, p. 1). Most argumentation schemes are accompanied by a 

list of so-called critical questions which users of the scheme can 

draw upon to anticipate the kind of objections that may be raised 

against such an argument. Argumentation schemes and their at-

tendant critical questions play such an important role in identify-

ing, analyzing, evaluating, and constructing arguments in natural 

language discourse that they have been widely studied in informal 

logic, rhetoric, and argumentation theory (Hastings 1963; Perel-

man and Olbrechts-Tyteca 1969; Kienpointner 1992; Walton 

1996; Grennan 1997; van Eemeren and Grootendorst 2004; God-

den and Walton 2007; Walton et al. 2008; Walton 2008a; Orsinger 

2011; Walton 2013a; Macagno 2015; Walton and Macagno 2015; 

etc.). Although this approach has also been criticized several times 

(for example, Pinto 2003; Lumer 2016; etc.), argumentation 

schemes have been employed in a wide range of domains and for 

many applications including legal reasoning (Walton 2005; Atkin-



A Network of Argumentation Schemes and Critical Questions 789 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

son and Bench-Capon 2021), decision support systems (van der 

Weide et al. 2006; Atkinson et al. 2008), ethical reasoning (Atkin-

son and Bench-Capon 2008a), argumentation framework (Atkin-

son and Bench-Capon 2008b; Gordon et al. 2007), argument inter-

change format (Rahwan et al. 2011), AI education (Green 2017), 

and even medical reasoning (Qassas et al. 2015).  

The list of critical questions accompanying an argumentation 

scheme helps the participants to easily answer the question, “what 

are the weaknesses of this argument?” One of the features of 

argumentation schemes that is key to evaluating whether an argu-

ment fitting a scheme should be judged as strong or weak is the list 

of associated critical questions—questions that can be asked (or 

assumptions that are held) by which a non-deductive argument 

based on a scheme might be judged to be (or presented as being) 

good or fallacious (Walton et al. 2008, pp.15-16). In fact, the set of 

critical questions accompanying a scheme allows for the vulnera-

bilities of an argument fitting the scheme to be identified. Thus, if 

a respondent identifies the argumentation scheme that the propo-

nent’s argument fits, then they can easily pinpoint its weakness 

using the list of critical questions attached to the scheme.  

After the weaknesses of the initial argument are identified, the 

second question can be raised: “how can I construct an argument 

that attacks or supports the former one?” The answer to the first 

key question may inspire the answer to this one. In other words, 

the participants can construct arguments that attack or support the 

former argument simply by backing up an answer to a critical 

question attached to the scheme that the former argument fits. In 

fact, an answer to the question and support for it together comprise 

an argument. The argument consists of an answer to the question 

as the conclusion and support for the answer as the premises.  

Furthermore, we know that argumentation schemes can be a 

powerful means for constructing arguments. We can easily con-

struct real arguments simply by substituting for the variables of an 

argumentation scheme. This means that argumentation schemes 

can be instantiated to construct an argument that backs up answers 

to a critical question. For example, we can apply the argumenta-

tion scheme from position to know to back up an answer to the 

trustworthiness question (“is expert E reliable as a source?”) of an 



790 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

argument from expert opinion as follows: “source A is in position 

to know about the expert E’s trustworthiness. Source A asserted 

that E is reliable. Therefore, E is reliable.” 

However, it is not the case that we can apply every argumenta-

tion scheme as support for an answer to a critical question since 

the application of a scheme partly depends on the semantic nature 

of the thesis to be proved. The semantic1 nature of a thesis is usu-

ally related to what kind of content the thesis contains rather than 

its logical form. While the logical nature of a thesis is related to its 

logical form, for example, whether the proposition that expresses 

the thesis is positive or negative, universal or particular, and so on, 

the semantic nature of a thesis is often related to its content, for 

example, whether it is about a property of things, a causal relation, 

an evaluation of something, or the desirability of an action, etc.2  

Not all of the semantic (material) relations that form the basis 

of the schemes can support all the possible conclusions or purpos-

es of an argument (Macagno 2015, p. 193). Therefore, we cannot 

arbitrarily apply argumentation schemes to back up an answer to a 

critical question. For example, we cannot apply the scheme for 

practical reasoning or the one for argument from waste to back up 

an answer to the interfering factor question: “are there other causal 

factors that could interfere with the production of the effect in the 

given case?” It is clear that these two schemes, if applied, would 

give irrelevant answers to the question. We can apply the schemes 

for practical reasoning and argument from waste only to assessing 

whether a course of action is desirable. Similarly, we can apply the 

scheme for argument from correlation to cause only to support a 

causal relation.  

In summary, only certain argument schemes can give a relevant 

answer to a critical question and what schemes are available to 

back up a particular answer depends on the semantic nature of the 

question. Furthermore, generally, the kind of answer to a critical 

 
1  The technical term ‘semantic’ is frequently used in Macagno (2015) and 

Macagno and Walton (2015) where they contrast it with the term ‘logical’ and 

classify and systemize clusters of argumentation schemes according to the 

semantic nature of the thesis to be proved.  
2 The classification of theses (theses about a property of things, a causal rela-

tion, an evaluation, or a proposal) is taken from Fahnestock and Secor (1982). 



A Network of Argumentation Schemes and Critical Questions 791 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

question that an arguer tries to back up does not influence their 

selection of suitable schemes, as will be demonstrated in the next 

section.  

However, while selecting the schemes that are likely to provide 

relevant answers to a critical question, we could not help but wan-

der about in the conventional scheme repertories. This is because 

schemes are not grouped according to critical questions. In the 

conventional scheme repertories, a list of critical questions is 

attached to each argumentation scheme, which is separate from the 

others. What is more, the critical questions are not followed by any 

scheme (see Fig. 2). In a nutshell, a scheme repertory is useful 

only for helping the participants determine an answer to the first 

key question, but not for the second key question. 

The purpose of this paper is to help users of argumentation 

schemes and critical questions find answers to both key questions. 

If we attach schemes to each critical question, critical questions 

can act as bridges between argumentation schemes. As a few 

critical questions are attached to a given argumentation scheme, so 

too can a few argumentation schemes be attached to a critical 

question. While a set of critical questions attached to a scheme 

helps identify the weakness of an argument fitting the scheme, a 

set of argumentation schemes attached to a critical question offers 

reasoning patterns that are suitable for use in justifying an answer 

to the critical question. Thus, an argumentation scheme can be 

connected through critical questions to the other schemes. 

From this perspective, we define a new notion, which we term 

“network of argumentation schemes and critical questions” 

(NASCQ). In a NASCQ, critical questions are grouped according 

to argumentation schemes and argumentation schemes are grouped 

according to critical questions. So, it serves to help argumentation 

participants find answers for both key questions. 

The rest of this paper is organized as follows: in section 2, we 

deeply scrutinize two kinds of possible answers to a critical ques-

tion. We also investigate whether this dichotomy restricts the 

selection of schemes appropriate to back up an answer to a critical 

question. The definition of a NASCQ is given in section 3. In 

section 4, we distinguish two important kinds of argumentation 

schemes: thesis-dependent and thesis-independent, which is useful 



792 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

for developing a NASCQ. Section 5 is dedicated to developing an 

example of a NASCQ. Its usefulness in constructing and evaluat-

ing complex arguments and in formal argumentation is briefly 

explored in sections 6 and 7.  

2. Two kinds of answers to a critical question 

A critical question has a number of possible answers, and thus, the 

kind of answer a participant is attempting to back up with evidence 

is important. This is because the set of argumentation schemes that 

can be applied to back up an answer to a critical question may be 

determined not only by the semantic nature of the question but 

also by the kind of answer an arguer tries to back up. 

Asking a critical question merely casts the argument into doubt 

and thus does not imply a stance. There may be two kinds of an-

swers to a critical question: positive and negative. In a nutshell, the 

positive answer to a critical question may be “yes,” while the 

negative may be “no.” For example, if the trustworthiness question 

of the scheme for argument from expert opinion—“is the expert 

reliable as a source?”—has been asked, then the answer “yes, they 

are reliable” is positive, while “no, they are not reliable” is nega-

tive. However, not all of the critical questions take the form of a 

yes/no question. Furthermore, even questions with the yes/no form 

do not all play the same role in argumentation. For example, the 

answer “yes” to the trustworthiness question of an argument from 

expert opinion plays the role of supporting the original argument, 

while, on the contrary, the same kind of answer to the other means 

question 3  of a practical reasoning argument or the third factor 

question of an argument from correlation to cause play the role of 

attacking the original argument fitting the scheme. Therefore, we 

can regard the positive answer as an answer that is favorable to the 

proponent and the negative answer as an answer that is favorable 

to the respondent. In other words, the positive answer to a critical 

question is one that supports an argument fitting the scheme it 

attaches to, while the negative answer is one that attacks an argu-

 
3  The other means question of practical reasoning is “[a]re there alternative 

possible actions to bringing about Si that could also lead to the goal?” (Walton 

et al. 2008, p. 96) 



A Network of Argumentation Schemes and Critical Questions 793 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ment fitting the scheme. Consequently, the positive answers can be 

taken as the premises of an argument fitting the scheme and the 

negative ones can be taken as rebuttals, or as undercutting or 

undermining the original argument. For another example, if the 

expertise question4 of an argument from expert opinion that does 

not take the yes/no form has been asked, then “they are a credible 

expert” and “they are not a credible expert” are respectively posi-

tive and negative answers to the argument.  

Given our definition, whether an answer to a critical question is 

positive or negative may be inconsistent with the sentence form of 

the answer. For example, a linguistically negative sentence “No, 

there is not a third factor” should be considered a positive answer 

to the third factor question.  

From the definition, backing up a negative answer to a critical 

question boils down to constructing an argument that attacks the 

former argument fitting the scheme to which the question is at-

tached, whereas backing up a positive answer to a question boils 

down to constructing a supporting argument.  

In remarkable contrast to the fact that the semantic nature of a 

critical question significantly restricts our selection of schemes, 

whether we try to back up a positive or negative answer to a criti-

cal question hardly restricts our selection of schemes. It just de-

termines which type of argumentation scheme we should employ. 

That is, if we use the positive form of an argumentation scheme 

for backing up a positive answer to a critical question, then, we 

may have to use the negative form for backing up a negative an-

swer. And, if we use the negative form of a scheme for backing up 

a positive answer to a critical question, then, we should use the 

positive form for backing up a negative answer. Moreover, many 

argumentation schemes, such as the schemes for argument from 

consequences and argument from values, have positive and nega-

tive forms. For example, we can not only establish the desirability 

of a course of action by instantiating the positive form of the ar-

gumentation scheme from consequences (argument from positive 

consequences) but also its undesirability by instantiating the nega-

 
4 The expertise question of argument from expert opinion is “[h]ow credible is E 

as an expert source?” (Walton et al. 2008, p. 310) 



794 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

tive form (argument from negative consequences). In Walton’s 

original exposition of argumentation schemes, some schemes do 

not have positive and negative forms; however, it is possible to 

conceive of two opposite forms for every scheme. For example, in 

the same way that the existence of a causal relation is drawn from 

the observation of a positive correlation between two events, the 

non-existence of a causal relation can be drawn from the claim that 

a positive correlation has not been observed. Therefore, the 

scheme from correlation to cause has positive and negative forms.  

In more general terms, whether a positive or negative answer 

follows from a scheme just depends on what we substitute for the 

variables of the scheme. That is, arguments that back up a negative 

answer to a question may be easily given by substituting negative 

(or sometimes positive) terms for the variables of an appropriate 

scheme. That is why Lumer (2016) criticizes Walton’s argumenta-

tion scheme approach for adding the variant “false” into many 

argumentation schemes, such as the scheme from expert opinion. 

According to him, it is superfluous to express both the positive and 

negative variants for every proposition in a scheme since a nega-

tive variant proposition can easily be created by substituting nega-

tive terms for a variable in a positive proposition. For example, the 

proposition “e asserts that proposition p is false” is equivalent to 

“e asserts that proposition non-p is true” in the scheme from expert 

opinion (Lumer 2016, p. 5).  

Notably, some schemes have critical questions about the exist-

ence of certain types of things, such as the other goals5 or other 

means question of the practical reasoning scheme or the third 

factor question of the scheme for argument from correlation to 

cause. For such questions, we can prove the existence of these 

things simply by providing an example; however, non-existence in 

an example doesn’t prove that things of that sort do not exist. For 

instance, suppose that the other goals question has been asked. The 

respondent can easily construct an argument that backs up the 

negative answer, “yes, there is one other goal G,” by instantiating 

a scheme, such as that for argument from values (see Fig. 4). 

 
5 The other goals question for the practical reasoning scheme is “do I have goals 

other than Si whose achievement is preferable and that should have priority?” 

(Walton et al. 2008, p. 96) 



A Network of Argumentation Schemes and Critical Questions 795 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

However, for the proponent, it is much more challenging to prove 

the positive answer, because denying the existence of the example 

goal G is insufficient to deny the existence of all possible goals. 

Therefore, for critical questions about the existence of something, 

it seems to us that different kinds of answers to a critical question 

cannot be dealt with simply by appealing to different forms (posi-

tive or negative) of a scheme or substituting different (positive or 

negative) terms for the variables of the scheme.  

Given this, our justification is twofold. First, we insist that the 

critical question about the existence of something should be more 

specific than the present one. In other words, such questions 

should have the form “is a particular thing…?” rather than “are 

there…?” For example, a more specific form of the other goals 

critical question may be “is the achievement of the goal G prefera-

ble to Si?” For this question, arguers can easily support its positive 

and negative answers by appealing to the positive or negative form 

of a scheme (argument from values in Fig. 4), respectively.  

Second, even in such cases, it is possible to back up the answer 

by applying the opposite form of a suitable argumentation scheme. 

In everyday argumentation, one of proponents’ usual tactics for 

dealing with critical questions about existence is to enumerate 

possible choices and then prove that all those choices are unavail-

able. In our example, the proponent can prove the non-existence of 

other goals by proving an argument “The possible goals other 

than my goal may be G1, G2, …, but none of them should be pur-

sued. Thus, there aren’t any other goals.” This can, in turn, be, 

supported by employing the negative form of the scheme from 

values (argument from negative values) several times. How many 

times we apply the scheme depends on the number of other possi-

ble goals. Dealing with critical questions about existence in such a 

way is a kind of argument from ignorance in everyday argumenta-

tion (Walton 1996; Walton et al. 2008) or negation as failure type 

reasoning, which is adopted for modeling the “closed world as-

sumption” in AI.  

As a result, we can say that the set of argumentation schemes 

that can be applied to back up answers to a critical question is 

determined only by the question, not by the kind of answer that 

needs to be backed up. This allows us to attach a set of schemes to 



796 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

a critical question no matter what kind of answer needs to be 

proven. Consequently, this allows us to define the concept of a 

NASCQ as we do in the next section. 

3. Definition of NASCQ 

A NASCQ is a constellation of schemes connected through critical 

questions. Different argumentation schemes and critical questions 

are interwoven together to form a NASCQ. It is a directed network 

that includes two kinds of nodes and two kinds of directed edges. 

The two kinds of nodes are the argumentation scheme node (AS-

node) and the critical question node (CQ-node). The two kinds of 

directed edges are the ask edge which goes from a CQ-node to an 

AS-node and the backup edge which goes from an AS-node to a 

CQ-node. There can be no edge that goes from an AS-node to 

another AS-node. There can be no edge that goes from a CQ-node 

to another CQ-node, either. In a NASCQ, one AS-node is connect-

ed to the other AS-nodes only through CQ-nodes. The figure 

below depicts a NASCQ that includes five schemes and eight 

critical questions. Rectangles represent schemes and rounded 

rectangles represent critical questions. In addition, dashed lines 

represent ask edges, while solid lines represent backup edges. As 

you can see, three critical questions (CQ-2, CQ-3, CQ-5) are 

connected to the argumentation scheme (Sch-1) through ask edges 

to point out the vulnerabilities of an argument fitting the scheme. 

The critical question (CQ-4) is connected to two schemes (Sch-1, 

Sch-3) through backup edges, which means that these schemes can 

be used to construct arguments that back up an answer to the 

question.  

 



A Network of Argumentation Schemes and Critical Questions 797 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

 
Fig. 1 A NASCQ 

 

A scheme repertory has two kinds of nodes, but only one kind of 

edge that goes from a CQ-node to an AS-node, and for this reason, 

it remains just a simple collection of isolated argumentation 

schemes. That is, in a scheme repertory, argumentation schemes 

are not connected with each other in any manner. The following 

figure depicts an example of scheme repertory that corresponds to 

the NASCQ depicted in Fig. 1.  

 

 
Fig. 2 An argumentation scheme repertory 

 

Unlike scheme repertories, in a NASCQ, argumentation schemes 

are connected with each other through the “bridges” of critical 

questions. A critical question, which is connected with a scheme 

through an ask edge, represents the vulnerabilities of an argument 

fitting the scheme and can in turn be used to evaluate the argu-

ment. And a scheme, which is connected with a critical question 

through a backup edge, can be used to construct arguments that 

back up an answer to the question. Therefore, a NASCQ helps 



798 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

argumentation participants find answers to the questions “what are 

the weaknesses of this argument?” and “how can I construct an 

argument that attacks or supports the former?” after an argumenta-

tion scheme that the former argument fits has been identified. 

It is also possible to imagine a NASCQ that involves only one 

kind of node and one kind of edge. In this kind of NASCQ, the 

nodes represent argumentation schemes and the edges stand for 

critical questions. Thus, every node is labeled with the name of a 

scheme, while every edge is labeled with the name of a critical 

question, as in Fig. 3 below.  

 

 
Fig. 3 A NASCQ with AS-nodes and CQ-edges  

 

The critical questions are represented by edges in Fig. 3, which 

may produce the misunderstanding that a critical question is given 

just as an answer justified only by means of a scheme. In real 

argumentation, participants often provide different answers to the 

same critical question and justify their answers by means of di-

verse schemes. This is why it is better for critical questions to be 

represented by nodes than by edges in a NASCQ. 

In order to build a real NASCQ, it is essential to determine how 

we can specify the schemes that can be applied to support an 



A Network of Argumentation Schemes and Critical Questions 799 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

answer to a critical question in a substantial way. In the next sec-

tion, a small part of the solution to this issue is addressed.  

4. Thesis-dependent and thesis-independent argumentation 

schemes 

As mentioned in the first section, proponents and respondents 

cannot arbitrarily apply argumentation schemes to support an 

answer to a critical question because choosing schemes usually 

depends upon the semantic nature of the thesis to be proved. But 

can we apply an argumentation scheme to support arbitrary critical 

questions? We found that, for some argumentation schemes, the 

answer is “yes.” but for the others, the answer is “no.” Some ar-

gumentation schemes can be applied to every critical question 

while the other schemes can be applied only to particular critical 

questions.  

In argumentation practice, the arguer’s choice of schemes de-

pends on the semantic nature of the thesis. For example, if a thesis 

is about the desirability of a course of action, we can apply the 

schemes for practical reasoning and those for argument from 

waste, consequences, established rules, need for help, danger 

appeal, distress, and threat. In contrast, for the same thesis, we 

cannot apply the schemes for argument from sign, abductive ar-

gument, argument from cause to effect, and argument from corre-

lation to cause because their conclusions are irrelevant to the 

desirability of a course of action. So, we call such schemes thesis-

dependent. 

However, most of the source-based argumentation schemes and 

the scheme for argument from analogy can be applied to any 

thesis, irrespective of their semantic nature. So, we call these 

thesis-independent schemes.  

A source-based scheme proves a thesis based on two factors: 

the presence of an agent that is in a position to know something 

related to the thesis, and the fact that the source has committed to 

the thesis. In most cases, there is no constraint on the kinds of 

theses to which a source is allowed to commit; a source can com-

mit to any theses. For example, an expert can give advice about 

what choice should be made or can present their opinion on what 

statement should be believed. For this reason, most source-based 



800 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

argumentation schemes can be applied to any kind of thesis. Note 

that there are some source-based argumentation schemes that we 

cannot include under the heading of thesis-independent schemes, 

such as the one for argument from witness testimony. Indeed, a 

witness—a person who has experienced an event such as a crime 

or an accident—is only allowed to state that something is true or 

false but is not allowed to assess the desirability of a course of 

action in a court. We classify the scheme for argument from anal-

ogy as thesis-independent also because there is no constraint on 

the semantic nature of conclusions that can be drawn from com-

parison between similar cases. Arguers can establish not only the 

desirability of a course of action but also the acceptability of a 

judgment based on an argument from analogy.6  

How can we decide whether an argumentation scheme is thesis-

dependent or thesis-independent? Unlike traditional logical infer-

ence patterns like modus ponens or syllogisms, which are highly 

abstract, contemporary argumentation schemes are based on their 

own semantic relations as well as logical relations. Argument from 

cause to effect, for example, is not only based on defeasible modus 

ponens, but also on a semantic causal relation between two events 

(Macagno 2015; Macagno and Walton 2015). An argumentation 

scheme may either be relevant to any kind of thesis or be relevant 

to only a certain kind of thesis according to the semantic relation it 

is based on. This requires us to thoroughly investigate the conclu-

sions that can be drawn from the application of a scheme in order 

to determine whether it is thesis-dependent or thesis-independent. 

Argument from position to know, for another example, can draw 

any kind of conclusion. There may be a person who is in a position 

to know what to believe, what causes what, what to do, what is 

good or bad, right or wrong, and any other kind of claim. The 

semantic relation between a source’s commitment to a proposition 

and the acceptability of the proposition is relevant to any kind of 

thesis. However, argument from sign, which is based on the se-

 
6 There have been many debates over the formulation of the scheme for argu-

ment from analogy. However, we say that, regardless its formulation, the 

analogical argumentation scheme is thesis-independent unless we set any 

constraint on possible conclusions that can be drawn from that kind of argu-

ment, as in the practical variant of the scheme.  



A Network of Argumentation Schemes and Critical Questions 801 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

mantic relation between a sign and an event, can only draw con-

clusions about the property of something or the occurrence of 

something, but it cannot draw any conclusions about whether a 

particular action is desirable (see Fig. 4). The semantic relation 

between a sign and an event is completely irrelevant to establish-

ing the desirability of bringing about a particular action.  

Schemes are basically classified as thesis-dependent or thesis-

independent according to the form of their conclusions. If any 

claim can be reduced to the form of the conclusion of a scheme, 

then it is thesis-independent otherwise the scheme is thesis-

dependent. We can easily see that no claim can be reduced to the 

form of the conclusion of the scheme from cause to effect (“there-

fore, in this case, B will occur”), the scheme from correlation to 

cause (“therefore, A causes B”), the practical reasoning scheme 

(“therefore, I ought (practically speaking) to carry out this action 

A”), the scheme from values (“V is a reason for retaining (retract-

ing) commitment to goal G”), or the scheme from classification 

(“A has property G.”). By substituting the variables of the conclu-

sions, we cannot get any kind of claims other than claims about the 

occurrence of something, a causal relation, the desirability of 

bringing about an action, the desirability of pursuing a goal, or the 

property of something, respectively (Walton et al. 2008). That is 

why those schemes are categorized as thesis-dependent. But it is 

self-evident that any claim can be reduced to the form of the con-

clusion of the schemes from analogy, expert opinion, position to 

know (“therefore, A is true.”), and thus, the schemes are catego-

rized as thesis-independent. However, this rule of thumb for cate-

gorization may not hold for some schemes, such as those of argu-

ment from sign or argument from witness testimony. Although, it 

seems like any kind of conclusion can be reduced to the form of 

the conclusions of argument from sign or witness testimony (“B is 

true in this situation,” “A may be plausibly taken to be true”), 

upon deeper inspection, as argued above, only certain kinds of 

conclusions can be drawn by applying those schemes (Walton et 

al. 2008). Therefore, to determine thesis-dependence or thesis-

independence, it is more accurate to carefully consider the conclu-

sions that can be drawn by applying a scheme.  



802 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Since most of the source-based argumentation schemes and the 

scheme for argument from analogy are thesis-independent, arguers 

can apply them to support answers to any critical questions. The-

sis-dependent schemes, on the other hand, can be applied to sup-

port answers to just a few critical questions. For example, argu-

ment from waste and argument from consequence can support an 

answer to the other means question of the scheme for practical 

reasoning. But they cannot provide a relevant answer to the possi-

bility question or the side effect question for the same scheme. We 

do not mean that a thesis-dependent scheme can be applied to 

support an answer to only one critical question. If a scheme is 

thesis-dependent, there exists at least one critical question, and any 

answers that are appropriate to answer it cannot be supported by 

employing the scheme. For example, argument from cause to 

effect is thesis-dependent, and it cannot be used to support an 

answer to any critical questions other than the possibility question 

and the side effect question in Fig. 4. 

So then, what schemes can be applied to support the answers to 

each critical question? In order to answer this question, we need to 

focus on the conclusion of every scheme. If the conclusion of a 

scheme can become an appropriate answer to a critical question, 

the scheme can be applied to support the answers to the question. 

Our careful scrutiny of the conclusions of Walton’s argumenta-

tion schemes (Walton et al. 2008) resulted in the following 

NASCQ. 

5. A potential NASCQ 

To begin with, we’d like to mention that this network is potential 

and incomplete. It is potential because all of its edges indicate 

possible relations between argumentation schemes and critical 

questions. In other words, it is not a model of a real or finished 

argumentation. It is also incomplete because it does not involve all 

of the recognized argumentation schemes and critical questions. A 

journal article is too limited in size to discuss all relations between 

them. So, we restrict the network to some of the relations with 

practical reasoning at the center. 

In this paper, we present a prototype of a NASCQ that was de-

veloped based on Walton et al.’s comprehensive work (2008). 



A Network of Argumentation Schemes and Critical Questions 803 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Building a full and complete NASCQ is not the aim of this paper. 

A larger NASCQ that involves other scholars’ schemes will be 

undertaken in the future. 

In this section, we stake the practical reasoning scheme as our 

starting point to build a network. A starting point is always re-

quired to create a network. We adopt the practical reasoning 

scheme as our starting point because it is very common and im-

portant in human reasoning. 7  Therefore, the application of a 

NASCQ is not restricted to practical argumentation. Rather, a 

NASCQ can be applied to a wide range of reasoning and argumen-

tation, including those about properties of things, causal relations, 

evaluations, or proposals. For example, the NASCQ can be used in 

causal argumentation since it includes causal argumentation 

schemes, such as argument from cause to effect, argument from 

sign, and argument from correlation to cause.  

In the figure below, solid arrows represent backup edges, while 

dashed arrows represent ask edges. A simple rectangle with a label 

inside it represents a thesis-dependent argumentation scheme and a 

snip diagonal corner rectangle with the label inside it represents a 

thesis-independent argumentation scheme. Finally, rounded grey 

rectangles with labels represent critical questions. In fact, snip 

diagonal corner rectangles should be connected with all the round-

ed rectangles since thesis-independent schemes can be used to 

construct arguments that back up any critical question. However, 

connecting every thesis-independent scheme with all critical ques-

tions makes the NASCQ very intricate and hard to understand. 

Therefore, in this NASCQ, we distinguish thesis-independent 

schemes from thesis-dependent schemes by using different rectan-

gle types.  

 
7 This does not mean that a NASCQ centered around the practical reasoning 

scheme should be different from one centered on another scheme. NASCQs 
may vary in their incomplete forms just because the schemes and question they 

involve may be different. However, regardless of which scheme we start from to 

build a NASCQ, the full and complete versions should mostly be the same, as 

long as they involve the same schemes and question. Furthermore, the complex 

relationship between argumentation schemes and critical questions is based on 

human real argumentation and is thus objective. 



804 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

 
Fig. 4 A potential NASCQ 



A Network of Argumentation Schemes and Critical Questions 805 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

 

It is redundant to present the full details of every argumentation 

scheme because it can be found in the compendium of schemes in 

Walton et al. (2008, pp. 308-346). 

What is important about the NASCQ is that we gave every crit-

ical question a name like we did in the earlier version of the 

scheme for practical reasoning (Walton et al. 2008, p. 96). Some 

critical questions attached to different schemes share the same 

name since they can be identified with each other with respect to 

their semantic nature. For example, CQ5 for practical reasoning, 

CQ3 for argument from need for help, CQ4 for argument from 

distress, and CQ3 for argument from consequences are very similar 

in the sense that they all inquire about the existence of a bad or 

undesirable consequence of bringing about a course of action. 

Therefore, they are given the same name, “side effect question,” in 

this paper.  

In addition, in Fig. 4, there are such critical questions as the in-

stance question or the conflict rule question that are not connected 

to any thesis-dependent argumentation schemes by a backup edge. 

The reasons for this are twofold. One is that an answer to such a 

critical question may be a factual claim whose truth should be 

assured by our sensory organs rather than by arguments. The other 

is that we failed to find any thesis-dependent argumentation 

schemes suitable for backing up the questions in the compendium 

of schemes of Walton et al. (2008). However, the questions can be 

backed up by any thesis-independent schemes. 

Below, we try to provide the details of Fig. 4, question by ques-

tion so that the readers can easily understand and utilize the 

NASCQ. We focus on which argumentation schemes can be ap-

plied to support the answers to a critical question. Herein lies the 

core of our paper. 

In what follows, to illustrate our approach, we adopt an imagi-

nary example of an instance of argumentation that concerns a 

larynx cancer case where several medical experts deliberate about 

the best treatment for a patient with early-stage superficial unilat-

eral larynx cancer, inspired by Qassas et al. (2015). Suppose that 

the argumentation starts from an argument fitting the practical 

reasoning scheme put forward by a surgeon: “Our goal is to pro-



806 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

vide the best cure rate. Hemi-laryngectomy is a means to achieve 

the goal. Therefore, we should perform a hemi-laryngecomy.” (A1) 

This will be used as a running example throughout this section.  

 

The other goals question is about the existence of a goal other 

than the one being considered. It is CQ1 for practical reasoning 

and an answer to it can be supported by applying the scheme for 

argument from values. This is because an argument from values 

takes a value judgment as a reason for retaining or retracting 

commitment to a goal. The question, “are there any other goals 

other than the best cure rate for larynx cancer?” is an example of 

an other goals question and a negative answer to it could be, “yes, 

there are; another goal could be to avoid any treatment risk.” Then, 

this kind of answer can be supported by instantiating the argumen-

tation scheme from values: “Avoiding any treatment risk will 

promote the value ‘life,’ so we should pursue the goal (thus, we 

should not perform the operation).”8 (A2) 

 

The other means question is CQ2 for practical reasoning. It asks 

whether there is another action that should be considered. We can 

support an answer to it by applying several argumentation 

schemes. Specifically, the schemes that establish whether a course 

of action is desirable by appealing to various factors, such as 

consequences, threat, rules, danger, need for help, distress, and 

waste. The other means question can even take the practical rea-

soning scheme as a means to back up its answers. In fact, the 

respondent may back up the other means question by making an 

appeal to another goal an agent should also pursue. In our exam-

 
8 The expression of this argument from values does not seem to accord with the 

standard form of the corresponding scheme in Walton et al.’s compendium 

(2008, p. 321), but we are sure that it follows the general pattern of an argument 

from values. Such an expression of an argument from values is inspired by the 

extended version of Atkinson and Bench-Capon’s practical reasoning scheme 
(2007, 2008). According to Atkinson and Bench-Capon, values denote some 

actual descriptive social attitude/interest that the proponent of an action may or 

may not wish to uphold or subscribe to. Those descriptions may be ‘life,’ 

‘liberty,’ ‘happiness,’ or others and not numeric quantities (2008, p. 138). The 

difference between the notions of value and goal can also be found in Atkinson 

and Bench-Capon (2008).  



A Network of Argumentation Schemes and Critical Questions 807 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ple, one of the other means questions may be “are there any alter-

native treatments that could be used instead of taking out the 

patient’s larynx?” And “no, there aren’t any possible alternatives” 

is the positive answer to it. We can support the positive answer to 

this typical question about existence as follows: “It may be possi-

ble to use radiotherapy or take out the patient’s larynx, but we 

cannot choose either since radiotherapy will lead to low voice 

quality and taking out the larynx will lead to voice loss. Therefore, 

there aren’t any alternative treatments” (A3). As can be seen, to 

support the answer to the question about existence, we adopt the 

argument type from ignorance or negation as failure type reason-

ing and employ the scheme for argument from consequences 

twice.  

 

The best means question is CQ3 for practical reasoning. It asks 

which is the most efficient among possible means; for example, “is 

hemi-laryngectomy the best among alternatives?" The best means 

should always be more efficient than any of the alternatives. Since 

an argument from alternatives recommends one of two candidate 

means, it is applicable to back up an answer to the best means 

question. It is also notable that if there are n alternatives to a pro-

posed action, an arguer should apply the scheme n times. We can 

support the positive answer to the best means question of our 

running example by using argument from alternatives as follows: 

“The possible alternatives are radiotherapy and taking out the 

larynx. Thus, either hemi-laryngectomy or radiotherapy can be 

selected. Radiotherapy is not our choice (because it fails to ensure 

voice quality). Thus, we should select hemi-laryngectomy. In 

addition, either hemi-laryngectomy or taking out the larynx can be 

selected, but we cannot choose to take out the larynx (because it 

will cause voice loss). Thus, we should, once again, select hemi-

laryngectomy. To sum up, hemi-laryngectomy is the best among 

alternatives.” (A4)  

 

The possibility question is shared by several argumentation 

schemes including practical reasoning (CQ4), argument from need 

for help (CQ2), argument from distress (CQ3), and argument from 

waste (CQ1). This question merely inquires whether a course of 



808 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

action is possible or not in a given circumstance. For example, we 

can ask “is it possible to perform the hemi-laryngectomy for the 

patient?” The arguers can support the possibility of bringing about 

an action under a given circumstance by applying the schemes for 

argument from cause to effect and argument from sign. In such a 

case, the arguers should mention a causal relation or symptomatic 

relation between a given circumstance and an action as a reason 

for the possibility. The arguers can argue “If B occurs, then it will 

become possible to bring about action A. B might occur. There-

fore, in this case, it is possible to bring about A” (argument from 

cause to effect). Or they can argue, “bringing about action A is 

possible when its sign B is true. B is true in this situation. There-

fore, it is possible to bring about A” (argument from sign). If we 

want to back up the negative answer to the example possibility 

question, we can construct an argument from cause to effect: “the 

patient’s old age is a cause that makes the hemi-laryngectomy 

impossible, and given that the patient is over 75, it is impossible to 

perform the operation” (A5). 

  

The side effect question is also connected with several argumen-

tation schemes. The schemes for practical reasoning (CQ5), argu-

ment from need for help (CQ3), argument from distress (CQ4), and 

argument from consequences (CQ3) share this question. It asks 

about the existence of a negative consequence of bringing about 

the action under consideration. “Would hemi-laryngectomy have 

any known negative consequences that ought to be taken into 

account?” can be a side effect question. We could not find the 

phrase “side effect” in the scheme for argument from consequenc-

es; however, “other opposite consequence” is used, and we believe 

they have the same meaning. The arguers can construct arguments 

that support an answer to this question by means of the schemes 

for argument from cause to effect and argument from sign. As you 

can see, the possibility question and the side effect question can be 

backed up by applying the same two argumentation schemes. This 

is because the side effect question is essentially about the possibil-

ity of a negative outcome. We can argue, “yes, there exists a side 

effect. That is to say, the patient might suffer from low voice quali-



A Network of Argumentation Schemes and Critical Questions 809 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ty because hemi-laryngectomy is one of the main causes of lower-

ing voice quality” (A6). 

  

The likelihood question is CQ1 for argument from consequences, 

and, simultaneously, CQ1 for argument from cause to effect and 

CQ1 for argument from sign. The point of these three critical 

questions is to determine the likelihood that one event will (may, 

must) occur given that the other event has occurred. More precise-

ly, CQ1 for argument from consequences is about the “likelihood 

that the cited consequences will (may, must) occur” (Walton et al. 

2008, p. 332) given that a course of action was brought about. CQ1 

for argument from cause to effect inquires about the strength of a 

causal generalization, in essence, the likelihood that the effect will 

occur given that the cause occurred. In addition, CQ1 of argument 

from sign is about “the strength of the correlation of the sign with 

the event signified” (Walton et al. 2008, p. 329)—that is, it is 

about the likelihood that an event occurred given that its sign is 

true. We unify these three questions under the heading of “likeli-

hood question.” Here, terms such as “likelihood” or “strength” 

have probabilistic meaning,9 and thus an answer to the question 

should have the form “x%.” This means that no thesis-dependent 

scheme can be applied to construct an argument that backs up the 

question except for the scheme for argument from random sample 

to population. In our running example, we have constructed an 

argument from cause to effect (A6) to support the existence of the 

side effect of hemi-laryngectomy. Now suppose that a likelihood 

question has been asked: “How strong is the causal generalization 

between hemi-laryngectomy and low voice quality?” We can 

respond, “according to a comprehensive study of a number of 

larynx cancer patients who have had a hemi-laryngectomy, only 

approximately 3% of them suffer from low voice quality. There-

fore, the likelihood is 3%—very, very low” (A7). 

 
9 From a mathematical standpoint, the term “possibility” also has a probabilistic 

meaning. However, the possibility question has the form of “is it possible?” not 

“what is the possibility?” and thus we cannot apply the scheme for argument 

from random sample to population. Instead, we can apply the schemes for 

argument from cause to effect and argument from sign.   



810 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

The evidence question is shared by two important schemes: 

argument from cause to effect (CQ2) and argument from conse-

quences (CQ2). We identify these two questions because both of 

them are about the existence of evidence that warrants the relation 

between a cause and an effect (argument from cause to effect) and 

an antecedent and a consequence (argument from consequences). 

The arguers can appeal to correlation or abduction to back up such 

a question. Some of the most important evidence one can have to 

argue for the existence of a causal relationship between two events 

is a correlation between them (Walton et al. 2008, p. 173). Fur-

thermore, the explanation of an abductive argument is, in most 

cases, a causal generalization, and that is why it can be said that “it 

might be better to express it (argument from effect to cause) ex-

plicitly as a species of abductive argument” (Walton et al. 2008, p. 

171). This means that the existence of a correlation between two 

events or the hypothesis that one causes another can be taken as 

evidence that supports a causal relation between them. In our 

example, we have tried to prove the impossibility of performing a 

hemi-laryngectomy by appealing to the causal relation between the 

patient’s age and treatment risk (A5). Thus, the evidence question 

asks “is there enough evidence to warrant the causal relation be-

tween the patient’s age and treatment risk?” We can then construct 

an abductive argument as follows: “patient A (who also suffers 

from larynx cancer but is different from the patient under discus-

sion) is very young and has undergone the hemi-laryngectomy. 

The operation was successful and there was no risk for patient A. 

However, patient B (who suffers from larynx cancer and is also 

different from the former two patients) has also had the hemi-

laryngectomy, but he was very old, 75. There was a risk for patient 

B. The best explanation for these facts is that old age is a cause of 

treatment risk” (A8). 

 

The interfering factor question is CQ3 for argument from cause 

to effect. The existence of an interfering factor means there is 

another causal relation between that factor and the absence of the 

effect cited in the original argument. Therefore, backing up an 

interfering factor question can be distilled to supporting another 

causal relation. This suggests that we can apply the schemes for 



A Network of Argumentation Schemes and Critical Questions 811 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

argument from correlation to cause and abductive argument to 

back up that question. For example, in order to argue for the exist-

ence of an interfering factor B that blocks the occurrence of A, we 

should support a causal relation between B and not-A; thus, we 

can construct arguments like “a positive correlation between B and 

not-A is observed; therefore, there exists a causal relation between 

B and not-A, and B is an interfering factor” or “the causal relation 

between B and not-A explains the collected data well; therefore, 

there exists a causal relation between B and not-A, and B is an 

interfering factor.” In the case of our running example, a respond-

ent may ask about the existence of an interfering factor in the 

causal relation between old age and treatment risk (A8). In this 

case, the patient’s good health can be such an interfering factor, 

the existence of which can be supported by means of an argument 

from correlation to cause: “there is a positive correlation between 

old but healthy larynx cancer patients and not having treatment 

risk; therefore, the patient’s good health is an interfering factor” 

(A9). 

 

The instance, conflict rule, and exception questions are the 

critical questions (CQ1, CQ2, and CQ3) for the scheme for argu-

ment from rules (Walton et al. 2008, p. 343). It is self-evident that 

we can support an answer to the exception question by applying 

the scheme for argument for exceptional case (Walton et al. 2008, 

p.344). 

 

The help question, which inquires whether the proposed action 

really helps someone else, is shared by the scheme for argument 

from need for help (CQ1) and argument from distress (CQ2).  

 

The distress question is CQ1 for the scheme for argument from 

distress and inquires whether someone is really in distress.  

 

The reassessment question is CQ2 for the scheme for argument 

from waste.  

 

The confirming question is the first in the list of critical questions 

for the scheme for argument from correlation to cause (CQ1). This 



812 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

question requires the arguer to confirm the real existence of two 

events.  

 

The third factor question is an important part of the scheme for 

argument from correlation to cause (CQ1). Since a third factor is 

one that causes both correlated things, providing support for an 

answer to the third factor question once again boils down to 

providing support for two other causal relations. For instance, 

suppose there exists a correlation between A and B, and the exist-

ence of a third factor C is revealed. Then, we need to support the 

causal relations between C and A and between C and B. Therefore, 

we can use the schemes for argument from correlation to cause 

and abductive argument to provide support for the answer to that 

question: “A positive correlation between C and A is observed, 

therefore C causes A. A positive correlation between C and B is 

observed, therefore C causes B. Eventually, C constitutes a third 

factor between A and B.” Note that we can find both backups and 

ask edges between the scheme for argument from correlation to 

cause and the third factor question. In this case, we are not going 

to commit the fallacy of circular reasoning or begging the question 

because we don’t appeal to the cited correlation but the other two 

correlations. For our example, let us imagine that the patient suf-

fers from severe abiosis (nutrition disorder) and a hoarse voice. 

The medical experts have seen many people that suffer from both 

abiosis and a hoarse voice. Thus, they may conclude that abiosis is 

a cause of hoarse voice. This is a fallacious use of the scheme for 

argument from correlation to cause (A10), and thus, the third factor 

critical question “is there a third factor that causes both abiosis and 

hoarse voice?” can be asked. We can also ask a more specific 

version of this question: “is larynx cancer a third factor that causes 

both the patient’s abiosis and their hoarse voice?” In such a case, 

the positive answer, “yes, larynx cancer is the third factor” can be 

backed up using the argument scheme from correlation to cause: 

“there is a positive correlation between larynx cancer and abiosis, 

thus larynx cancer is a cause of abiosis. And larynx cancer is also 

a cause of hoarse voice because there is a positive correlation 

between larynx cancer and hoarse voice. Therefore, larynx cancer 

is a third factor” (A11). 



A Network of Argumentation Schemes and Critical Questions 813 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

The coincidence question asks whether there exists any reason 

that establishes that the cited correlation is any more than a coinci-

dence. This question is CQ3 for the scheme for argument from 

correlation to cause. In order to disprove that the cited correlation 

is mere coincidence, we should prove that the correlation is a real 

causal relation, and this means that we may be able to apply the 

schemes for argument from correlation to cause and abductive 

argument to back up the answer to the coincidence question. How-

ever, the argumentation scheme from correlation to cause cannot 

be used as such since it causes us to commit the fallacy of circular 

reasoning or begging the question. In our example, we have con-

structed an argument by appealing to the correlation between old 

but healthy larynx cancer patients and the absence of treatment 

risk from a hemi-laryngectomy (A9). Then, the coincidence ques-

tion “Is there any reason to think that the correlation is any more 

than a coincidence?” (Walton et al. 2008, p. 329) can be asked. In 

order to support the negative answer to this question, we can con-

struct an abductive argument whose conclusion indicates that there 

is no causal relation: “Patient A (who also suffers from larynx 

cancer but is different from the patient under discussion) was not 

healthy and underwent the hemi-laryngectomy. The operation was 

successful and there was no risk to patient A. However, patient B 

(who suffers from larynx cancer and is also different from the 

former two patients) also had a hemi-laryngectomy, but he was 

very healthy. There was a risk for patient B. The best explanation 

for these facts is that patients’ good health does not cause treat-

ment risk, and thus the observed correlation is mere coincidence” 

(A12). 

 

The other event question, which is CQ2 for the scheme for argu-

ment from sign, asks about the existence of “other events that 

would more reliably account for the sign” (Walton et al. 2008, p. 

329) Here, the other event that can account for the sign under 

consideration may be a cause of the sign, and thus, we can prove 

answers to this question by constructing an argument using the 

correlation to cause or abductive argument schemes. Let us as-

sume that the medical experts who oppose the hemi-laryngectomy 

procedure need to prove that the patient’s health is not good. To do 



814 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

so, they provide an argument from sign: “his pulse rate is really 

high, thus, he isn’t in good condition” (A13). This argument is 

based on the implicit generalization that high pulse rate is a sign of 

poor health. In this case, the following is an appropriate other 

event question: “he exercised just a few minutes ago; does exer-

cise more reliably account for his high pulse rate?” Answers to this 

question can easily be backed up using an argument that appeals to 

the correlation between exercise and high pulse rate (A14).  

 

The property question, which inquires whether something defi-

nitely has the cited property, can be backed up using several ar-

gumentation schemes, including argument from sign, argument 

from witness testimony, argument from division, argument from 

composition, argument from verbal classification, argument from 

definition to verbal classification, argument from perception, and 

argument from memory. This critical question can be found in the 

list attached to the scheme for argument from verbal classification 

(CQ1). In fact, Macagno, in his means-end classification of 

schemes, selected argument from sign, argument from verbal 

classification, argument from composition, and argument from 

division as the schemes that can be used to support the attribution 

of a factual property (2015, p. 197). Furthermore, we can apply the 

schemes for argument from composition or division to back up an 

answer to the property question because those schemes prove that 

something has a specific property based on the relation between 

the whole and its parts. We connect the scheme for argument from 

witness testimony to the property question (through a backup 

edge) since a witness, as a person who has experienced something 

happening, can attribute a property rather than making claims 

about the desirability of a course of action or a causal relation. For 

our example, suppose that, in order to prove that the patient is not 

in good condition, a medical expert constructs an argument from 

verbal classification: “he is (should be classified as) an old man, 

thus he is not in good condition” (A15). Then, “is he really old?” 

can be an archetypal property question. The expert can back up the 

positive answer to the question by constructing an argument from 

definition to verbal classification: “he is 75, which is over 60. So, 



A Network of Argumentation Schemes and Critical Questions 815 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

he is old” (A16). This answer is based on the definition that one 

who is over 60 is old. 

 

The hereditary question is about the divisional or compositional 

hereditariness of a cited property with regard to a certain aggre-

gate. Argument from composition and argument from division 

share this question, since they are all based on the relation between 

the whole and its parts.  

 

The difference and other case questions are CQ1 and CQ3 of the 

scheme for argument from analogy, respectively. The difference 

question is about the existence of a difference between two cases, 

while the other case question is about the existence of another case 

that is also similar to the former case and in which the cited propo-

sition is false (Walton et al. 2008, p. 315). Since difference or 

similarity between two cases should be considered with regard to a 

certain property, if two cases share a property, we say that they are 

similar with respect to the property, and in contrast, if two cases 

have different properties, then we say that they are different with 

regard to those properties. Therefore, for the purpose of backing 

up answers to these questions, we can apply all argumentation 

schemes available for supporting the property question. 

  

The truth question merely inquires whether the cited proposition 

is true in a given case. This question can be found in the list for the 

scheme for argument from analogy (CQ2).  

 

The assumption question is CQ2 for the scheme for argument 

from verbal classification, which inquires whether the cited verbal 

classification is merely “based on an assumption about word usage 

that is subject to doubt” (Walton et al. 2008, p. 319). 

 

The adequate definition and stipulative or biased definition 

questions apply to the scheme for argument from definition to 

verbal classification.  

 

The progress, burden of proof, and proof strength questions 

apply to the scheme for argument from ignorance. 



816 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

The poll and doubt questions apply to argument from popular 

opinion. 

 

The action and ground questions apply to the scheme for argu-

ment from popular practice.  

 

The expertise, field, opinion, trustworthiness, consistency, and 

backup evidence questions are the six critical questions for the 

scheme for argument from expert opinion. Other source-based 

arguments, like argument from position to know and argument 

from witness testimony, have some of these questions as can be 

seen in the figure.   

 

To sum up, Fig. 4 involves 29 argumentation schemes and 39 

critical questions. Among the 29 argumentation schemes, 6 

schemes are thesis-independent, and the rest are thesis-dependent 

and can be applied only to specified critical questions. Further-

more, among the 39 critical questions, 18 have more than one 

thesis-dependent scheme that can be used to back up their answers, 

and the rest have no such schemes. We also put forward 16 argu-

ments to illustrate the usefulness of our NASCQ. 

6. NASCQ and complex arguments 

A NASCQ can be used to construct and reconstruct, and then, in 

turn, evaluate complex arguments. Constructing and reconstruct-

ing complex arguments is one of the issues that has received sig-

nificant attention in argumentation theory and informal logic 

(Freeman 1991, 2011; Hitchcock 2017; Walton 2006; Henkemans 

1992; van Eemeren and Grootendorst 2004; van Eemeran et al. 

2014; etc.). Hitchcock (2017) defined the notion of complex ar-

gument in terms of subordinate and superordinate argument. A 

subordinate argument is “an argument whose conclusion is a 

premiss of another argument” (Hitchcock 2017, p. 455) while a 

superordinate argument is “an argument with a premiss that is the 

conclusion of another argument.” (Hitchcock 2017,p. 455) There-

fore, a complex argument is a chain of simple arguments, that is, 

“a set of two or more simple arguments each of which is either 

superordinate to or subordinate to at least one other argument in 



A Network of Argumentation Schemes and Critical Questions 817 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

the set, and one of which (the main argument) is not subordinate to 

any other argument in the set” (Hitchcock 2017, p. 455). 

Since every scheme is useful for constructing and reconstruct-

ing simple arguments and a complex argument is a chain of simple 

arguments, we should also construct or reconstruct complex argu-

ments by means of a chain of argumentation schemes. We call 

such a chain of argumentation schemes, which captures the inter-

nal structure of a complex argument, a scheme chain. But how can 

such a scheme chain be created? How should we connect schemes 

in such a scheme chain? Generally, if a chain is to be strong, every 

link should be connected with another proper link, not an arbitrary 

one. Similarly, in a scheme chain, every scheme should be con-

nected with another scheme that is suitable for providing it with a 

proper premise.  

In a complex argument, the conclusion constitutes a premise for 

another argument. Therefore, it is self-evident that a scheme chain 

should have a scheme that is supported by another scheme in-

volved in the chain, that is, a scheme whose premise is supported 

by another scheme involved in it.  

The most important thing is that the positive answer to a critical 

question attached to a scheme can be counted as a premise in that 

scheme. For example, Walton et al. (2008) proposed some new 

versions of the scheme for argument from expert opinion whose 

critical questions convert into conditional premises (Version Ⅱ 

and Ⅲ) (Walton et al. 2008, pp. 19-20) and, more remarkably, a 

version where every critical question is converted into an individ-

ual premise (Version Ⅳ) (Walton et al. 2008, p. 20). This demon-

strates that the positive answer to a critical question is, in its es-

sence, a premise of the scheme to which the question is attached. 

Furthermore, based on the fact that the positive answer to a critical 

question attached to an argumentation scheme can be a premise of 

that scheme, Gordon et al. managed to model critical questions in 

their Carneades argumentation system (2007, pp. 887-888). More 

interestingly, every real and ordinary premise of a scheme can be 

the positive answer to a critical question attached to the scheme, 

since it is possible for the scheme to have a critical question that 



818 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

merely asks whether a premise of the scheme holds 10  (Verheij 

2003). For instance, the major premise of the scheme for argument 

from expert opinion is the positive answer to the field question, 

whereas the minor premise is the positive answer to the opinion 

question.  

To sum up, the answers to all of the critical questions attached 

to a scheme can be regarded as premises of the scheme. Moreover, 

in a scheme chain, which captures the internal structure of a com-

plex argument, a scheme is connected to another scheme through a 

statement that is a premise of one scheme and the conclusion of 

the other. Therefore, it seems that in a scheme chain, a scheme is 

connected to another scheme through a critical question whose 

positive answer becomes a premise of one scheme and the conclu-

sion of the other. Finally, a scheme chain can be defined as a set of 

two or more schemes, each of which has a critical question that is 

connected to another scheme or is connected with a critical ques-

tion of another scheme and one that (the main scheme) is not 

connected with any critical question for any other scheme.11 This 

shows that a scheme chain may, in fact, be part of a NASCQ. In 

other words, if all the schemes and critical questions involved in a 

scheme chain are also included in a NASCQ, all the connections 

between schemes and questions (both ask and backup edges) 

 
10 Such a critical question, which is used merely to question whether a premise 

of a scheme holds, is a question whose positive answer becomes Gordon et al.’s 

(2007) ordinary premise. This kind of critical question was first identified by 

Verheij (2003). Verheij noted that this kind of critical question is redundant, but 

Gordon et al. argued that it is necessary in a dialogue setting in order to distin-

guish different kinds of premises, including ordinary premises, and claimed that 

Verheij’s account can only “make sense in a logical setting” (Gordon 2007, p. 

879). It seems that, in this paper, we follow Gordon et al.’s (2007) account. 

Although the role of critical questions which merely ask whether a premise of a 

scheme holds cannot be neglected, such questions are not attached to some of 
the schemes in Walton et al.’s (2008) compendium, for example, the practical 

reasoning scheme. Therefore, we argue that critical questions of this kind 

should be attached to every scheme to build a more comprehensive NASCQ 

capable of capturing more complex arguments.   
11  This definition follows the pattern of Hitchcock’s (2017) definition of a 

complex argument. 



A Network of Argumentation Schemes and Critical Questions 819 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

involved in the chain should also be included in the NASCQ.12 

Consequently, a NASCQ can be used to construct and reconstruct 

complex arguments. Every part of a NASCQ, which contains at 

least one critical question node and at least two scheme nodes, 

represents the pattern of reasoning to which a complex argument 

may conform. The positive answer to the critical question repre-

sented by a CQ-node becomes the conclusion of an argument and 

a premise of another in a complex argument structure. 

For example, let us consider the following complex argument: 

“We should perform a hemi-laryngectomy for the patient in order 

to cure his larynx cancer. He is in good condition.” This complex 

argument is an instantiation of the scheme chain represented in 

Fig. 4: Practical reasoning-Possibility question-Argument from 

cause to effect. Such identification makes it possible to supply the 

missing parts of the complex argument and, in turn, to evaluate it. 

The structure of this complex argument can be analyzed by means 

of the scheme chain as follows (solid rectangles represent explicit 

propositions, while dashed ones represent implicit propositions):  

 
Fig. 5. Reconstructing complex arguments with a NASCQ 

 
12 The basic assumption behind this claim is that the NASCQ is ideally compre-

hensive; that is, it involves all possible connections that may exist between 

schemes and questions (both ask and backup edges) in real argumentation.  



820 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

As shown in the figure above, the positive answer to the possibil-

ity question of practical reasoning (it is possible to perform a 

hemi-laryngectomy) becomes the proposition that connects two 

arguments involved in the complex argument; viz., it is the con-

clusion of one argument and a premise of the other. Such analysis 

carried out by means of the scheme chain teaches us that the com-

plex argument has two missing premises and a missing conclusion. 

After identifying a scheme chain or a part of a NASCQ that cap-

tures the structure of a complex argument, the critical questions of 

all schemes involved in the chain can be asked, except for the 

question that connects the schemes.13 In such a way, we can evalu-

ate complex arguments in a dialectical setting as we normally do 

with simple arguments using argumentation schemes. Not only can 

we fill in the missing parts of a complex argument, but we can also 

construct complex arguments using a NASCQ.  

In a dialectical setting, the proponent, who tries to back up the 

positive answer to a critical question asked by the respondent, in 

fact constructs a complex argument whose ultimate conclusion 

represents the thesis they are attempting to prove. Constructing 

complex arguments by means of a NASCQ without having asked 

any critical questions is proleptic in nature. To construct a com-

plex argument, first the superordinate argument, which contains 

the ultimate conclusion, must be constructed using a specific 

argumentation scheme. Then, the proponent should construct a 

subordinate argument whose conclusion is a premise of the super-

ordinate argument. The proponent should select a question from 

the critical questions attached to the scheme that the superordinate 

argument fits, whose positive answer is the premise they would 

support. Later, the proponent can construct a subordinate argument 

by applying a scheme associated with that critical question in the 

NASCQ. In such a way, the proponent can construct complex 

arguments by “walking along the road” of a NASCQ. This can be 

seen as proleptic since the proponent constructs subordinate argu-

ments in anticipation of the critical questions the respondent will 

 
13  Merely asking the critical question that connects two parts of a complex 

argument is redundant because an answer to the question is already provided. 

However, we think it should not be seen as unreasonable to present another 

argument in favor of or against the answer provided.  



A Network of Argumentation Schemes and Critical Questions 821 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ask (for more details about proleptic argumentation, see Walton 

2008b). 

7. NASCQ and formal argumentation 

Recently, the field of AI has witnessed growing research interest 

in formal argumentation due to its explanatory power and its ca-

pacity to handle inconsistency. Most formal argumentation mech-

anisms are based on Dung’s (1995) landmark work on abstract 

argumentation framework (AF), which consists of a set of argu-

ments and the binary attack relations between them. Semantics is 

used as a criterion for identifying acceptable arguments and draw-

ing plausible conclusions. A set of arguments identified by seman-

tics is called an extension.  

It has been recognized that employing argumentation schemes 

in formal argumentation systems will help artificial agents im-

prove their reasoning capability (Bex et al. 2003; Verheij 2003). 

Some attempts have been made to combine argumentation 

schemes with formal AFs (Gordon et al. 2007; Prakken 2010; van 

Gijzel and Prakken 2012; Atkinson and Bench-Capon 2008b). The 

basic idea behind those attempts is that we can capture the internal 

structure of arguments involved in an AF by means of argumenta-

tion schemes. Furthermore, it is also possible to characterize the 

relation between arguments as that of attack or support by means 

of critical questions.  

An AF might be bipolar depending on whether it involves both 

attack and support relations or only attack relations (Cayrol and 

Lagasquie-Schiex 2013; Cohen et al. 2014). It has been shown that 

there exists both attack and support relations between arguments in 

everyday argumentation (Polberg and Hunter 2018). However, 

most argumentation semantics are defined in terms of attack, and 

bipolar AF should be converted into an equivalent mono-polar AF 

to apply semantics. If an AF allows for only simple arguments, the 

AF should be bipolar to consider all possible attacks and support 

relations between the arguments. If an AF allows complex argu-

ments, the AF can consider all possible attack and support rela-

tions because any support relation can be included in a complex 

argument structure (Dung 1995). Since our NASCQ approach 



822 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

makes it possible to reconstruct complex arguments, it can be 

applied not only to bipolar AFs but also to mono-polar AFs.  

The NASCQ approach can, first of all, be adopted in argumen-

tation-based decision support systems. In addition to getting justi-

fied answers, there is another benefit from modeling arguments 

and attacks with argumentation schemes and critical questions in 

Afs—such modeling can provide us with suggestions for exploring 

the critical questions that have not been assigned as an answer in 

order to make hidden assumptions explicit or to gather additional 

information about the case under discussion (Qassas et al. 2015, p. 

288). Suppose that the argument “his pulse rate is really high, 

thus, he isn’t in good condition” (A13 in the example of section 5) 

is justified by an AF. The framework not only supplies the output 

proposition that the patient is not in good condition but also sug-

gests looking into the likelihood of a causal relation existing be-

tween high pulse rate and poor health or looking into whether 

there are other events that could more reliably account for the high 

pulse rate because the other event and likelihood critical questions 

are attached to the scheme for argument from sign. Next, since in 

our NASCQ, the other event question is connected to the schemes 

for argument from correlation to cause and abductive argument, 

the framework can also suggest that additional information be 

gathered about the correlation between the observed other event 

and the sign or additional information that indicates which is better 

explained by the other event. Modeling every argument and attack 

with schemes and critical questions may also be associated with 

Shi et al.’s argument-based modal belief logic, which is based on a 

so-called argumentation-support frame, where every argument and 

attack is labelled by a set of propositions that it supports or attacks 

(2018).  

Of course, it may be impossible to model every argument and 

attack involved in an AF by means of schemes and critical ques-

tions. However, if an analyzer manages to do so, then no recog-

nized connection between schemes and critical questions should 

be arbitrary; instead, it should appear in a NASCQ, assuming that 

the NASCQ is complete—that is, it involves all the schemes and 

questions observed in the framework and all possible relations 

(both asks edges and backup edges) among them. For example, in 



A Network of Argumentation Schemes and Critical Questions 823 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

section 5, we built an AF which consists of 16 arguments and the 

attack or support relations among them. The figure below, wherein 

solid arrows represent attacks and dashed arrows represent sup-

ports, depicts the AF.  

 

 
Fig. 6. NASCQ and AF 

 

As shown above, every attack or support relation between two 

arguments is a relation between two schemes, which are bridged 

by means of a critical question in our NASCQ. For this reason, we 

can adopt a NASCQ for the sake of constructing arguments and 

establishing the attack or support relations among them. However, 

it should be noted that in order to make use of a NASCQ in such a 

way, one must ensure that it includes as many argumentation 

schemes and critical questions and possible relations as possible. 

A NASCQ can also be used to improve argumentation in multi-

agent systems (or in dialogue games). Argumentation schemes, as 

dialectical tools, can be employed in multi-agent systems where 

agents try to construct, analyze, and evaluate arguments. Some 

systems are capable of utilizing a single argumentation scheme 

(e.g. Atkinson and Bench-Capon 2007, 2008a) or a set of schemes 

(Wells 2014; Panisson et al. 2021). Now, it is time to introduce a 

NASCQ for multi-agent systems. Human arguers can construct 

arguments by ‘walking along the road’ of a NASCQ, and so can 

artificial agents, who are not faced with the problem of memoriz-



824 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ing a large and complicated NASCQ. Furthermore, even if the 

argumentation schemes of a NASCQ are repeatedly modelled, 

they do not give rise to circular arguments (Walton and Batten 

1984).  

Panisson et al. (2021) have devised an argumentation scheme-

based argumentation framework that determines the acceptability 

of an argument on the basis of considering attendant critical ques-

tions in addition to Dung’s (1995) admissible semantics. They say 

that an argument, instantiated from an argumentation scheme, can 

be an acceptable instance of that scheme only when all critical 

questions related to the scheme are answered positively (Panisson 

et al. 2021, Definition. 3, p. 364). Positively answering (or nega-

tively answering when the agent wants to attack the argument) 

critical questions should result in the construction of further argu-

ments that support the positive (or negative) answer to the ques-

tion, and as a result, chained arguments are needed (Panisson et al. 

2018, 2021). Then, a NASCQ can show the agents what reasoning 

patterns they are able to instantiate to positively (or negatively) 

answer a critical question. 

In multi-agent systems, agents’ construction of arguments 

should be based on dynamic, not static, knowledge bases. There-

fore, agents’ deliberation may shift to information-seeking where 

relevant facts are brought into play (Walton 2013a). A NASCQ 

can also be a useful tool for information-seeking and for updating 

an agent’s knowledge base. Let us assume that an agent has con-

structed an argument by applying an argumentation scheme. In 

order to establish the acceptability of the argument, the agent 

should positively answer all of the attendant critical questions. 

However, if positive answers to the critical questions are not con-

tained in the agent’s knowledge base, then the agent may have to 

search for some relevant information. In such a case, a NASCQ, in 

which a set of argumentation schemes are attached to a critical 

question, will be of great help. For example, if an agent tries to 

seek information that is relevant to back up the positive answer to 

the side effect question of practical reasoning, then the associated 

schemes (argument from cause to effect, argument from sign, and 

all thesis-independent schemes) provide hints about what kind of 

evidence should be gathered: the cause of the side effect is not 



A Network of Argumentation Schemes and Critical Questions 825 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

given, there is no sign of the side effect, someone who is in posi-

tion to know about the side effect said that there would be no such 

a side effect, or the side effect didn’t occur in similar situations 

and cases.  

One of the most ambitious and challenging projects of AI is ar-

gumentation mining, which is closely related to formal argumenta-

tion, computational linguistics, natural language processing, and 

other AI fields. Argumentation mining has been defined as the 

automated detection of the argumentation structure and the classi-

fication of both its component elements and their argumentative 

relationships (Moens 2018). The detection of argumentation 

schemes in natural language discourse is a crucial portion of the 

spectrum of argumentation mining because “argumentation 

schemes provide categories of arguments, together with criteria for 

recognizing them” (Macagno 2021, p. 294). 

By utilizing a NASCQ, we are able to narrow down the scope 

of the search for schemes that correspond to an argument. It 

should be determined to which critical question in the NASCQ the 

conclusion of the argument applies. Then, of the argumentation 

schemes attendant to the critical question, the most appropriate 

one should be selected by considering the premises of the argu-

ment. For example, if the conclusion of an argument can be count-

ed as an answer to the side effect question, then the instantiated 

argumentation scheme will be the scheme for argument from cause 

to effect, or argument from sign, or one of the thesis-independent 

schemes (assuming that the NASCQ in Fig. 4 is complete). The 

more argumentation schemes and critical questions a NASCQ 

involves, the more reliable the scheme detection engine we can 

build.  

As the definition above shows, argumentation mining also in-

cludes the detection of relationships between arguments beyond 

the detection of schemes. It seems to us that revealing relation-

ships between arguments is a relatively easy job because if one 

argument attacks/supports another, then they have a contradicto-

ry/common proposition. However, in everyday argumentation, an 

attack/support relation may exist between arguments even though 

they do not have any contradictory/common propositions. Consid-

er the two arguments below (taken from our example).  



826 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Argument 1. A hemi-laryngectomy is a means to cure larynx can-

cer, so we should perform a hemi-laryngectomy on the patient. 

 

Argument 2. The patient might suffer from low voice quality be-

cause a hemi-laryngectomy is one of the main causes of lowering 

voice quality. 

 

It is obvious that the latter argument attacks the former, but they 

do not have any contradictory conclusions. A NASCQ can be of 

some help in a situation like this to detect attack/support relation-

ships between arguments that have no contradictory/common 

propositions. After clearly identifying the argumentation schemes 

of two arguments, common critical questions (that appear both in 

the list of critical questions attached to the suspected attack-

er’s/supporter’s scheme [through ask edges] and the list of critical 

questions that the suspected attacker/supporter’s scheme is at-

tached to [through backup edges]) should be detected. If the con-

clusion of the suspected attacker/supporter can be reduced to a 

negative/positive answer to one of those common questions, then 

it can be declared that the arguments are in an attack/support 

relation. In the above example, the former argument is an instance 

of practical reasoning, and the latter is an instance of argument 

from cause to effect. In our NASCQ, practical reasoning is con-

nected to 5 critical questions through ask edges, and argument 

from cause to effect is connected to 2 questions through backup 

edges. We can see that the possibility and side effect questions are 

shared by those two schemes. Furthermore, the conclusion of the 

latter argument is an instance of the negative answer to the side 

effect critical question of the former argument. Thus, it can be 

determined that the latter argument attacks the former. Here we 

should once again note that the more argumentation schemes and 

critical questions a NASCQ involves, the more reliable the argu-

mentative relationship detection engine we can build.  

8. Conclusions and future work 

In this paper, we defined the new concept of a NASCQ. A 

NASCQ is a constellation of schemes, each of which is connected 

with another through the “bridge” of a critical question. Therefore, 



A Network of Argumentation Schemes and Critical Questions 827 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

two kinds of nodes and two kinds of edges are possible in a 

NASCQ. AS-nodes represent argumentation schemes, while CQ-

nodes represent critical questions. The two kinds of edges are ask 

edges which go from a CQ-node to an AS-node and backup edges 

which go from an AS-node to a CQ-node. The critical question at 

the starting point of an ask edge can be asked against an argument 

fitting the scheme at the endpoint of the edge. Furthermore, the 

scheme at the starting point of a backup edge is applicable to back 

up an answer to the critical question at the endpoint of the edge. 

For example, in Fig. 4, five critical questions (the other goals, 

other means, best means, possibility, and side effect questions) are 

connected with the scheme for practical reasoning through ask 

edges, and this means that these questions can be asked against an 

argument fitting the scheme. Two schemes (abductive argument 

and argument from correlation to cause) are connected to the side 

effect question through backup edges, and this means that these 

schemes can be instantiated to construct an argument that backs up 

an answer to the question.  

One of the important contributions of this paper is the distinc-

tion between thesis-dependent schemes and thesis-independent 

schemes. Thesis-independent schemes are applicable to back up 

answers to any critical questions, while thesis-dependent schemes 

are applicable only to certain questions. Such a distinction makes 

it easier to find possible candidate schemes to back up an answer 

to a critical question. Our distinction between thesis-dependent 

and thesis-independent argumentation schemes may also contrib-

ute to classifying a system of schemes.  

As mentioned in the introduction, the NASCQ approach will be 

particularly useful for constructing an argument that attacks or 

supports the former argument. Once a proponent proves their own 

standpoint by putting forward an argument that supports the stand-

point, the main concern of the proponent and the respondent is 

then to construct an argument that supports or attacks the former 

argument, respectively. To that end, both of them first identify the 

argumentation scheme that the former argument fits. Next, they try 

to construct an argument whose conclusion is the positive or nega-

tive answer to one of the critical questions attached to the scheme. 

Then, they can instantiate one of the argumentation schemes con-



828 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

nected to the critical question through ask edges in the NASCQ. 

By “walking along the road” of the NASCQ, both parties can 

proceed with their debate. We also showed that the NASCQ ap-

proach is instrumental in constructing or reconstructing complex 

arguments and formal argumentation frameworks.   

In order to skillfully make use of a NASCQ, the users should 

memorize not only scheme-question connections but also question-

scheme ones. This seems to vitiate the approach suggested in 

Hansen and Cohen’s (2011) learner efficiency criterion for evalu-

ating methods of informal logic. However, from another perspec-

tive, the NASCQ approach may increase learner efficiency since it 

provides a method for classifying and categorizing clusters of 

argumentation schemes and critical questions. How fast one mas-

ters the art of using schemes depends on the way they are classi-

fied and categorized as well as their number. As mentioned above, 

our approach makes it possible to unify similar critical questions 

that appear in different schemes and, in turn, reduce the number of 

critical questions. Furthermore, argumentation schemes can be 

grouped according to critical questions to which they can be ap-

plied, and this provides an easier way to memorize many schemes. 

For example, the schemes for practical reasoning, argument from 

consequences, need for help, distress, and waste are all connected 

to the other means question and thus connected to the possibility 

and side effect questions. This approach also has a broader scope 

since it is applicable to complex arguments as well as simple ones.  

If it is going to serve the purpose of helping students and argu-

mentation theorists construct and evaluate arguments, a NASCQ 

should be stored in a computer program.14 Suppose that an argu-

ment is presented as the starting point of argumentation in a class 

on critical thinking or argumentation theory. The teacher can tell 

the students to ask some critical questions to evaluate the argu-

ment. A student may think “what questions should I ask?” In the 

 
14 In this paper, we do not explain the structural and algorithmic details of this 

kind of program not only because of space constraints, but also because it is 

beyond our aim. The aim of this paper is just to provide the theoretical ground 

for programming for such pedagogical purposes. Accordingly, we will turn our 

attention to developing such software based on a formal definition of NASCQ 

in future work.  



A Network of Argumentation Schemes and Critical Questions 829 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

case that the questions don’t come easily to their mind, they can 

identify the scheme of the argument and click on its name on the 

program interface. The program will then display a set of possible 

critical questions attached to the scheme. The students can then 

ask proper questions in reference to the set. When an answer is 

given to a question, the teacher can ask “how can you justify your 

answer?” If a student is not sure how to craft an argument for their 

answer, they can click on the name of the question asked. Then, 

the program will display a set of possible argumentation schemes 

attached to the critical question. By instantiating any of these 

schemes, they can support the answer. Through such exercises, 

students can improve their skills for evaluating and constructing 

arguments. 

Some students may ask an appropriate critical question that has 

not been stored in the program. In this case, the set of critical 

questions stored in the database should be updated. Similarly, a 

student may apply any new scheme to justification. Then the set of 

argumentation schemes should be updated. As a result, NASCQ 

will become more complete. 

Our ambition is to build a NASCQ that involves all of the ar-

gumentation schemes that have been defined in the literature, but 

this is highly challenging since all possible relations between 

many argumentation schemes and critical questions should be 

carefully taken into account. Furthermore, argumentation schemes 

haven’t been proposed by a single scholar but by many scholars, 

and each scholar provided their own definitions and classifications 

of the argumentation schemes.  

Our future research task is to introduce a complete NASCQ into 

computer programs for teaching critical thinking and argumenta-

tion theory and various AI tools.  

References 

Atkinson, K. and T. Bench-Capon. 2007. Practical reasoning as pre-

sumptive argumentation using action-based alternating transition sys-
tems. Artificial Intelligence 171(10-15): 855-874. 

Atkinson, K. and T. Bench-Capon. 2008a. Addressing moral problems 

through practical reasoning. Journal of Applied Logic 6: 135–151. 



830 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Atkinson, K. and T. Bench-Capon. 2008b. Abstract argumentation 
scheme frameworks. In AIMSA 2008, LNAI 5253, eds. D. Dochev, M. 

Pistore and P. Traverso, 220-234. Berlin/Heidelberg: Springer-

Verlag. 
Atkinson, K. and T. Bench-Capon. 2021. Argumentation schemes in AI 

and law. Argument & Computation 12: 417-434. 

Atkinson, K., T. Bench-Capon and S. Modgil. 2008. Argumentation for 
decision support. In DEXA 2006, LNCS 4080, eds. S. Bressan, J. 

Küng and R. Wagner, 822-831. Berlin/Heidelberg: Springer-Verlag. 

Bex, F., H. Prakken, C. Reed, D. Walton. 2003. Towards a formal ac-

count of reasoning with evidence: Argumentation schemes and gen-
eralizations, Artificial Intelligence and Law 11(2-3): 125-165. 

Cayrol, C. and M. Lagasquie-Schiex. 2013. Bipolarity in argumentation 

graphs: Towards a better understanding. International Journal of Ap-
proximate Reasoning 54(7): 876-899. 

Cohen, A., S. Gottifredi, A. J. García, G. R. Simari. 2014. A survey of 

different approaches to support in argumentation systems. The 

Knowledge Engineering Review 29: 513-550. 
Dung, P. M. 1995. On the acceptability of arguments and its fundamen-

tal role in nonmonotonic reasoning, logic programming, and n-person 

games. Artificial Intelligence 77(2): 321-357. 
Eemeren, F. H. van and R. Grootendorst. 2004. A systematic theory of 

argumentation. Cambridge: Cambridge University Press. 

Eemeren, F. H. van and B. Garrsen. 2014. Argumentation by analogy in 
stereotypical argumentative patterns. In Systematic approaches to ar-

gument by analogy, eds. H. J. Ribeiro, 41-56. Heidelberg/New 

York/Dortrecht/London: Springer. 

Eemeren, F. H. van, B. Garssen, E. C. W. Krabbe, A. F. S. Henkemans, 
B. Verheij and J. H. M. Wagemans. 2014. Handbook of argumenta-

tion theory. Dordrecht/Heidelberg: Springer Reference. 

Fahnestock, J. and M. Secor. 1982. A Rhetoric of argument (2nd ed). 
Boston: McGraw Hill Publisher. 

Freeman, J. B. 1991. Dialectics and the macrostructure of arguments. A 

theory of argument structure. Berlin/New York: Foris. 
Freeman, J. B. 1995. The appeal to popularity and presumption by 

common knowledge. In Fallacies: Classical and contemporary read-

ings, eds. H. V. Hansen and R.C. Pinto, 263-273. University Park: 

Pennsylvania State University Press. 
Freeman, J. B. 2011. Argument structure. Representation and theory. 

Dordrecht/New York: Springer. 



A Network of Argumentation Schemes and Critical Questions 831 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Gijzel, B. van and H. Prakken. 2012. Relating Carneades with abstract 
argumentation via the ASPIC+ framework for structured argumenta-

tion. Argument and Computation 3(1): 21-47. 

Godden, D. M. and D. Walton. 2007. Advances in the theory of argu-
mentation schemes and critical questions. Informal Logic 27(3): 267-

292. 

Gordon, T. F., H. Prakken and D. Walton. 2007. The Carneades model 
of argument and burden of proof. Artificial Intelligence 171: 875-

896. 

Green, N. L. 2017. Argumentation scheme-based argument generation to 

support feedback in educational argument modeling systems. Interna-
tional Journal of Artificial Intelligence Education 27: 515-533. 

Grennan, W. S. 1997. Informal logic. London: McGill-Queen’s Univer-

sity Press. 
Hansen, H. V. and D. H. Cohen. 2011. Are there methods of informal 

logic? In Argumentation: Cognition and community. Proceedings of 

the 9th international conference of the Ontario Society for the Study 

of Argumentation (OSSA), ed. F. Zenker, 1-13. Windsor, ON: OSSA. 
Hastings, A. 1963. A reformulation of the modes of reasoning in argu-

mentation. Ph.D. dissertation. Evanston, Illinois: Northwestern Uni-

versity. 
Henkemans, S. A. F. 1992. Analysing complex argumentation. The 

reconstruction of multiple and coordinatively compound argumenta-

tion in a critical discussion. Amsterdam: Sic Sat. 
Hitchcock, David. 2017. Informal logic and the concept of argument. In 

On reasoning and argument. Essays in informal logic and on critical 

thinking. Argumentation library 30, ed. F. H. van Eemeren, 447-476. 

Springer International Publishing AG.  
Jovicic, T. 2002. Authority-based argumentative strategies. Doctoral 

dissertation in the department of theoretical philosophy. Uppsala, 

Sweden: Uppsala University.  
Juthe, A. 2005. Argument by analogy. Argumentation 19: 1-27. 

Kienpointner, M. 1992. How to classify arguments. In Argumentation 

illuminated, eds. F. H. van Eemeren, R. Grootendorst, J. A. Blair and 
C. A. Willard, 178-188. Dordrecht: Foris. 

Lumer, C. 2016. Walton’s argumentation schemes. In Argumentation, 

objectivity, and bias: Proceedings of the 11th international confer-

ence of the Ontario Society for the Study of Argumentation (OSSA), 
eds. P. Bondy and L. Benacquista, 1-20. Windsor, ON: OSSA.  

Macagno, F. 2015. A means-end classification of argumentation 

schemes. In Reflections on theoretical issues in argumentation theo-



832 Sung-Jun Pyon and Yong-Sok Ri 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

ry. Argumentation library 28, eds. F. H. van Eemeren and B. Gars-
sen, 183-201. Springer. DOI:10.1007/978-3-319-21103-9_14 

Macagno, F. 2021. Argumentation schemes in AI: A literature review. 

Introduction to the special issue. Argument & Computation 12: 287-
302. 

Macagno, F. and D. Walton. 2015. Classifying the patterns of natural 

arguments. Philosophy and Rhetoric 48(1): 26-53. 
Moens, M.-F. 2018. Argumentation mining: How can a machine acquire 

common sense and world knowledge? Argument & Computation 9: 

1-14. 

Orsinger, R. 2011. The role of reasoning in constructing a persuasive 
argument. San Antonio: San Antonio Office. 

Panisson, A. R., A. Ali, P. McBurney and R. H. Bordini. 2018. Argu-

mentation schemes for data access control. In Computational Models 
of Argument (COMMA), 361-368. 

Panisson, A. R., P. McBurney and R. H. B. Bordini. 2021. A computa-

tional model of argumentation schemes for multi-agent systems. Ar-

gument & Computation 12: 357-395.  
Perelman, C. and L. Olbrechts-Tyteca. 1969. The new rhetoric. Notre 

Dame: University of Notre Dame Press. 

Pinto, R. C. 2003. Commentary on C. Reed and D. Walton ‘Argumenta-
tion schemes in argument-as-process and argument-as-product. In In-

formal Logic at 25: Proceedings of the Windsor Conference, eds. J. 

A. Blair, D. Farr, H. V. Hansen, R. H. Johnson and C. W. Tindale. 
Windsor, ON: OSSA.  

Polberg, S. and A. Hunter. 2018. Empirical evaluation of abstract argu-

mentation: Supporting the need for bipolar and probabilistic ap-

proaches. International Journal of Approximate Reasoning 93: 487–
543. 

Prakken, H. 2010. An abstract framework for argumentation with struc-

tured arguments. Argument and Computation 1(2): 93-124. 
Qassas, M. A., D. Fogli, M. Giacomin and G. Guida. 2015. Analysis of 

clinical discussions based on argumentation schemes. Procedia Com-

puter Science 64: 282-289. 
Rahwan, I., B. Banihashemi, C. Reed, D. Walton and S. Abdallah. 2011. 

Representing and classifying arguments on the semantic web. The 

Knowledge Engineering Review 26(04): 487-511. 

Reed, C. and G. Rowe. 2004. Araucaria: software for argument analysis, 
diagramming and representation. International Journal of AI Tools 

13(4): 961-980. 



A Network of Argumentation Schemes and Critical Questions 833 

 

© Sung-Jun Pyon and Yong-Sok Ri. Informal Logic, Vol. 42, No. 4 (2022), pp. 787–833. 

Shi, C., S. Smets, and F. R. Velázquez-Quesada. 2018. Beliefs supported 
by binary arguments. Journal of Applied Non-Classical Logics 28(2-

3): 165-188.  

Verheij, B. 2003. Dialectical argumentation with argumentation 
schemes: An approach to legal logic. Artificial Intelligence and Law 

11(2-3): 167-195. 

Walton, D. 1996. Argumentation schemes for presumptive reasoning. 
Mahwah, NJ: Lawrence Erlbaum Associates. 

Walton, D. 1997. Appeal to expert opinion. University Park: Penn State 

University Press. 

Walton, D. 2005. Argumentation methods for artificial intelligence in 
law. Berlin/Heidelberg: Springer. 

Walton, D. 2006. Fundamentals of critical argumentation. Cambridge: 

Cambridge University Press. 
Walton, D. 2008a. Informal logic. New York: Cambridge University 

Press. 

Walton, D. 2008b. Proleptic argumentation. Argumentation & Advocacy 

44: 143-154. 
Walton, D. 2010. Similarity, precedent and argument from analogy. 

Artificial Intelligence and Law 18(3): 217-246. 

Walton, D. 2013a. Methods of argumentation. New York: Cambridge 
University Press. 

Walton, D. 2013b. On a razor’s edge: evaluating arguments from expert 

opinion. Argument & Computation 5(2-3): 139-159. 
Walton, D and F. Macagno. 2015. A classification system for argumen-

tation schemes. Argument & Computation 6(3): 219-245. 

Walton, D., C. Reed and F. Macagno. 2008. Argumentation schemes. 

Cambridge: Cambridge University Press. 
Walton, D. and M. Koszowy. 2016. Arguments from authority and 

expert opinion in computational argumentation systems. AI and soci-

ety 32(4): 483-496.  
Weide, T. L. van der, F. Dignum, J. Ch. Meyer, H. Prakken and G. A. 

W. Vreeswijk. 2006. Personality-based practical reasoning. In 

ArgMas 2008, LNAI 5384, eds. I. Rahwan and P. Moraitis, 3-18. Ber-
lin/Heidelberg: Springer-Verlag. 

Walton, D. and L. M. Batten. 1984. Games, graphs and circular argu-

ments. Logique et Analyse 27(106): 133-164. 

Weitzenfeld, J. 1984. Valid reasoning from analogy. Philosophy of 
Science 51(1): 137-149. 

Wells, S. 2014. Supporting argumentation schemes in argumentative 

dialogue games. Studies in Logic, Rhetoric and Grammar 36(49): 
171-191.