Patterns of Responses to Abusive Ad Hominem Attacks: The Case of Facebook News-commenting Adam Karalama Safsatasına Verilen Yanıtların Sav Modelleri: Facebook Haber Yorumlarından Örnekler

News-commenting in social media is a platform that offers an opportunity for online deliberation through argumentative discussion. Yet aggressive exchanges between commenters with clashing ideologies have become a prominent feature of online news-commenting. One example of the aggressive exchanges is the use of direct personal attacks among news-commenters, namely abusive ad hominem. In this paper, I aim to reveal the patterns of responses given to abusive ad hominem attacks by studying the comments to news items in Facebook. The reason is to shed light on how the discussion evolves after the ad hominem attack. The patterns that this paper illustrates are a summary of studying the responses to 20 ad hominem attacks that figure in the comments to news items topicalizing various social problems in Turkey. The examples were drawn from those that topicalise ‘violence against women in Turkey’. Three patterns were identified: (1) abusive ad hominem as a response to an abusive ad hominem attack; (2) refusing to carry on the discussion; and (3) critically evaluating the abusive ad hominem attack. These patterns show that the pragmadialectical definition of the ad hominem fallacy proves to be functional in understanding its role in blocking the way to the resolution.


Introduction
As online interaction forms have become increasingly popular, internet users make wide use of the opportunities to comment on the news they read and interact with other users. News items that draw on social problems (such as gender inequality, racism, assault, and violence) are controversial, and they have a potential to attract commenters from varying viewpoints or ideologies. People can easily reach these news items from the newspapers' social media pages and leave their comments about the news they read and at times have discussions with other users in the same platform. News-commenting in social media may involve an argumentative exchange in which a commenter expresses a standpoint concerning the news he/she has read and uses arguments to defend this standpoint once he faces critical responses from other commenters. Despite the fact that this platform offers an opportunity for online deliberation through an argumentative discussion, aggressive exchanges between commenters based on clashing ideologies have become a prominent feature of online newscommenting 1 . It is gradually turning into a medium of hostile interactions between the commenters rather than a constructive medium for exchanging viewpoints on controversial topics. One example of the aggressive exchanges is the use of direct personal attacks among news-commenters. A direct personal attack, namely abusive ad hominem, is generally defined in argumentation theory as attacking a person's character or personality instead of targeting his arguments. For this reason, there is a dominant tendency to regard them as an instance of fallacious reasoning (van Eemeren & Grootendorst, 2004; van Eemeren, Garssen, & Meuffels, 2009;Walton, 1987). In this paper, I aim to provide some patterns of responses given to ad hominem attacks made during commenting for news items in Facebook and shed light on how a direct personal attack affects the discussion.
A number of approaches in argumentation theory provided a description of an ad hominem attack. For instance, Walton (1987), states that ad hominem attacks are those "which criticize another argument by questioning the personal circumstances or personal trustworthiness of the arguer who advanced it" (p.317). In accordance with their theory of "dialogue types", Woods and Walton (1989) note that ad hominem abusive need not always be considered fallacious. When an arguer lacks the relevant expertise or authority to make an argument, then questioning the trustworthiness of the arguer could be regarded as reasonable.
In pragma-dialectical approach to argumentation (van Eemeren & Grootendorst, 2004; van Eemeren, 2010), a direct personal attack, namely abusive ad hominem, is one that is directed to the personality of the opponent and aimed at discrediting the opponent party as a serious discussion partner. The attack involves a negative characterization of the opposing party in terms of his/her character, abilities, skills, etc., rather than addressing his arguments ( van Eemeren, 2010; van Eemeren, Garssen, & Meuffels, 2012). Ad hominem attacks are always regarded as fallacious 2 argumentative moves in Pragma-Dialectics as they violate the first rule of a critical discussion. 3 The rule reads as follows: "Discussants may not prevent each other from advancing standpoints or from calling standpoints into question" (van Eemeren & Grootendorst, 2004, p. 190). This rule assigns to the protagonist of a standpoint the right to express his/her standpoint freely and to the antagonist the right to call a standpoint into question freely. When a discussant violates this rule in the confrontation stage of a discussionthe stage in which the difference of opinion between the parties in the discussion becomes clear -1 The aggressive and hostile behaviour online is widely addressed in the literature with the name 'flaming' and a number of studies discussed its role and prevalence in online discussion settings (see for example O'Sullivan, & Flanagin, 2003;Hutchens, Cicchirillo, & Hmielowski, 2015). 2 For an empirical account of regarding ad hominem as fallacious in Pragma-Dialectics, see van Eemeren, Garssen, & Meuffels, 2009).
3 Van Eemeren and Grootendorst (2004) came up with ten commandments for reasonable discussants, each corresponding to a rule of a critical discussion. These are; (1) freedom rule: parties should have the freedom to advance and criticize a standpoint, (2) obligation-to-defend rule: a party who raised a standpoint should defend that standpoint if he/she is asked to do so, (3) standpoint rule: attacks should bear on a standpoint that has actually been raised, (4) relevance rule: a standpoint should be defended by relevant argumentation, (5) unexpressed premise rule: discussants should not falsely attribute unexpressed premises to each other, (6) starting-point rule: discussants should not falsely present something as an accepted starting point, (7) validity rule: arguments used to defend a standpoint should be valid, (8) argument scheme rule: parties should use appropriate argument schemes to defend a standpoint conclusively, (9) concluding rule: a conclusively defended standpoint may not receive further doubts and an inconclusively defended standpoint may not be maintained, and (10) language use rule: parties should use appropriate language in defending their standpoints. the discussant commits the fallacy of ad hominem, an argumentative move that hinders the process of resolving a difference of opinion. A potential example of abusive ad hominem is as follows: "No, I will not reply. I see no need to defend my views against the objections of ignoramuses." ("Ad hominem", n.d.). In this example, the arguer rules out the opponents as serious discussion partners by calling them "ignoramuses", and therefore, blocks the way to a possible resolution of the dispute.
By adopting the pragma-dialectical definition of fallaciousness concerning ad hominem attacks, in this paper, I aim to specify the patterns of responses given to ad hominem attacks in the argumentative context of news-commenting in Facebook and discuss how abusive ad hominem affects the argumentative exchange. Two research questions are posed to this end: (1) how does a direct personal attack affect the discussion; and (2) what do the dialectical profiles of responses to abusive ad hominem attacks look like? The dialectical profiles serve to summarize the patterns of responses given to abusive ad hominem attacks. They are instrumental in evaluating the nature of the argumentative moves made in reaction to abusive ad hominem attacks. They also show that such attacks are, at all times, regarded as unreasonable moves in the discussion, and thus, affect the rest of the discussion accordingly. The patterns of responses to ad hominem attacks that this paper illustrates rest upon observing the comments (i.e., which involve abusive ad hominem) to 20 news items topicalizing various political events and social problems in Turkey. These news items were published between 2016 and 2018 on the Facebook pages of two mainstream newspapers in Turkey: Hürriyet and Milliyet. For the purpose of illustration, I will draw my examples from the news items that topicalize 'violence against women in Turkey'.

News-commenting in Facebook
Online news-commenting is available in various forms. One is the official internet pages of the newspapers which allot a separate section for the comments for each news item. An important feature of these comments is that they are subject to prior inspection by the editors/moderators of these pages. The comments that include offence or defamation against people and institutions, the ones that contain evidently false statements, or the ones that may constitute a crime are eliminated during this pre-publishing process in line with the rules of the newspaper (netiquette) or the rules of a higher-order inspection mechanism 4 . Because this prior inspection of the comments in terms of appropriateness takes time and does not guarantee that all comments will be published, many internet users opt for commenting on news via the newspapers' social media pages rather than their official internet pages, which gives them an opportunity to react to the news they read instantly and involve in discussion with other commenters in a rather easy way.
Popular social media networks such as Facebook and Twitter are recognized as participatory tools (Polat, 2005) which facilitate rapid and effective information sharing, expression of different views, and opinion-formation. These features make social media networks fruitful platforms for the public to participate in deliberation. A number of studies have addressed the deliberative function of social media networks (Halpern & Gibbs, 2013;Kim, 2011;Lewiński, 2010;Steenkamp & Hayde-Clarke, 2014;Vesnic-Alujevic, 2012;You, Lee, Kang, & Go, 2015). Commenting on news items that point at social problems is a means to get involved in deliberation aimed at opinionformation in the social media. Some news items that topicalize social problems such as violence against women, wrongdoings of people in authority, actions against human and animal rights, or restrictions of freedom of speech attract a wide variety of comments from internet users with different ideologies. These comments help form a pool of public opinion. Because the power of social media is well-recognized in our present day, the public opinion that is collected from social media platforms may well be used to inform or affect political decision-making processes.
Facebook is one of the most widely used social media networks which enable its users to instantly comment on the news they read. People can easily follow the news from the Facebook pages of the newspapers, leave their comments instantly, and have discussion with other commenters about the content of the news in question. Despite the fact that Facebook is a platform that may facilitate fruitful deliberative discussion on controversial news items, the opportunity to post comments in a spontaneous and easy way in Facebook makes it possible for its users to leave comments that include personal attacks or insulting expressions against other users. Impolite and offensive behavior is a well-recognized phenomenon in the literature of online news commenting (Neurauter-Kessels, 2011;Rowe, 2015;Santana, 2014;Upadhyay, 2010). The fact that there are offensive comments, however, does not mean that Facebook is free of any rules and regulations about proper behavior in online posting.
The first constraint imposed upon the users of Facebook about their posts including news-comments is its Terms of Service (2015), which every user has to agree with before he/she can start using the social networking site. The safety conditions of the Terms urge the users to undertake commitments against, for instance, bullying, intimidating, or harassing other users, posting content that includes hate speech, threat, and discrimination. Facebook holds the authority to remove any content or information the users post if they believe it violates the statements in the Terms. This means that any comment that infringes these conditions can be removed by Facebook.
Along with the constraints of Facebook, there are also specifications by the newspapers themselves about 'proper behaviour' in their Facebook pages. The Turkish newspaper Hürriyet, for instance, have page rules that all the internet users following the newspaper from Facebook are expected to abide by. Publishing comments or posts that contain swearwords and insulting expressions, or ones that attack the rights of people or institutions may lead to the post being removed by the page administrators.
Although comments to newspapers' Facebook pages are subject to the above stated regulations, abiding by these regulations depends mostly upon users' self-commitments to the proper online behaviour standards rather than a strict mechanism that monitors the appropriateness of the contents posted simultaneously. Most comments that contain personal attacks accompanied by swearwords and insulting expressions can remain long time in Facebook before they are removed due to being found inappropriate or being reported as abuse.
Commenting on a news item in Facebook is possible in two options. One is the 'comment' option which involves commenting directly on the news published, and the other is the 'reply' option which allows for commenting on or replying to the comments of other people. While direct comments to a news article address mostly the content or people involved in the news item itself, commenting via 'reply' option creates a possibility to set up a separate discussion in which people exchange views about a given comment. Although it is also possible in direct commenting option to have interactive discussions by referring to the names or words of certain other commenters, commenting by using the 'reply' option creates an easier ground to personalize the conversation and enhance interactivity compared to the former commenting option, which often includes independent comments that do not address a specific user. Direct comments appear hierarchically higher than reply-comments as the latter ones are 'comments to comments'. Henceforth, I will call the former type of comments 'first-order comments', and the latter 'second-order comments' to maintain a clearer cut distinction while drawing on examples. The following is a depiction of how first and second-order comments appear in Facebook:

Figure 1. The representation of first and second-order comments in Facebook
First-order comments for a news article are not necessarily made to start a discussion with other users. However, once they are posted, they have a potential to attract reactions from other users and initiate a A first-order comment Second-order comments discussion in the second-order commenting area. To put it differently, if the comments are directed heavily against a certain user`s comment rather than the content of the news item, the discussion often continues in the second-order commenting area.
While studying the responses to ad hominem attacks, I will draw on the responses to ad hominem attacks in the second-order commenting area. More precisely, in this paper, I will focus on a form of interaction in which a person (Commenter A) leaves a first-order comment expressing a standpoint related with the content of a news item. Later, this commenter receives an abusive ad hominem attack from another person (Commenter B) in the second-order commenting area. Finally, Commenter A gives a response to this ad hominem attack. The scheme of the interaction is as follows: News item X (i) Commenter A: Standpoint expressed concerning a news item (first-order comment) (ii) Commenter B: Directing an abusive ad hominem attack against A (second-order comment) (iii) Commenter A: Response to the abusive ad hominem attack (second-order comment) In the following section, i will shed light on how the discussion evolves after an abusive ad hominem attack, as it is illustrated in step (iii) above and identify the patterns of responses given to such an attack.

Patterns of Responses Given to Abusive Ad Hominem Attacks in News-commenting
The patterns identified in this study rest upon working through the 20 responses given to ad hominem attacks in the comments to various news items. These news items were published on the Facebook pages of the newspapers Hürriyet and Milliyet between 2016 and 2018. The topics of the news items ranged from politics to various social events and problems. For the sake of brevity and conciseness, and in order to ensure the thematic unity of this paper, examples will be drawn from only one news topic, namely 'violence against women in Turkey', which offers possibility to examine the three patterns of responses to abusive ad hominem attacks that this paper draws on. These patterns of responses are as follows: (1) abusive ad hominem as a response to an abusive ad hominem attack; (2) refusing to carry on the discussion; and (3) critically evaluating the abusive ad hominem attack. These patterns will be exemplified with three news items and responses to these news items. The remaining 17 news items that were used as the data of the study in arriving at these three patterns were treated only as numerical data. All the data that were analysed corresponded to one of these three patterns. Because the topics of most news items were varied and lacked thematic unity, the examples that this paper draws on are confined to the news items that offer this unity, namely about 'violence against women'.
Before moving on to illustrate these patterns by drawing on examples, it is relevant to mention some limitations of this study. For one thing, this study does not aim to make general or cultural claims about how Turkish people respond to ad hominem attacks, nor does it maintain that these three patterns are the only patterns that can figure in various argumentative contexts of news-commenting. Such an objective would require the analyst to carry out a quantitative empirical study that draws on a wide range of data including various online interaction settings and a wider time span. This study, on the other hand, is mainly designed as a qualitative study that attempts to shed light on the prominent argumentative tendencies in responding to ad hominem attacks based on a given period of time and a specific social media domain, namely Facebook.
Another limitation concerns the availability of data matching the model of interaction provided above. Finding data that illustrate an argumentative exchange in which the attacked one comes back to the discussion to give reaction to the attacker is not an easy task. Therefore, a selective sampling procedure was adopted in the data collection. Nevertheless, tough limited, 20 cases that were examined already provide us with a good starting point to see how an ad hominem attack shapes the discussion. The following sections will treat the aforementioned three patterns of responses to ad hominem attacks individually.

Abusive ad hominem Attack as a Response to an Abusive ad hominem Attack
In this pattern of response, a first-order commenter, as the protagonist of a standpoint, faces an ad hominem attack as a reply and responds to this move with another ad hominem attack. The following is the dialectical profile 5 depicting these successive argumentative moves.
A: +/p B: -/p because A is so-and-so ?/(+/p) because A is so-and-so A: maintain +/p because maintain +/p because B is so-and-so B is so-and-so The dialectical profile above suggests that the arguer A puts forward, in the first-order commenting area, a positive standpoint with regard to the proposition p underlying the news item. This standpoint is attacked by another arguer, B, with an ad hominem move, which suggests that +/p is not acceptable on the grounds that A is so-and-so. An alternative to this dialectical profile is the one in which A puts forward a negative standpoint (-/p) concerning the proposition underlying the news item. In this case, the opponent participant B, who raises the ad hominem attack against A, aims to argue that -/p is not acceptable on the grounds again that A is so and so. For the sake of homogeneity, I will show the dialectical profiles based on one type of standpoint, namely +/p, but the examples I will be drawing on may also illustrate a case in which A holds the standpoint -/p.
In the dialectical profile above, by attacking A personally, B shifts the focus of attention from the propositional content of A's standpoint to the personal characteristics of A, represented by the expression "so-andso". B's ad hominem attack triggers another ad hominem attack as a response, this time from A, who has put forward the standpoint. A tries to maintain his/her standpoint by attacking the character of B. The discussion stops after the last personal attack. The difference of opinion can potentially be mixed or non-mixed 6 in such an exchange, but a mixed difference of opinion (i.e., in which the discussion partners hold the opposing standpoints) is the predominant one. The first pattern of response, illustrated by Dialectical Profile (1) is the most common pattern in the data observed. 14 of the total 20 cases examined display such a profile of response to an abusive ad hominem attack. This first pattern of response to an ad hominem attack can be exemplified with the following case: On February 23, 2016 a news item titled "Noone heard Cansel shouting for help" was posted to the Facebook page of the newspaper Hürriyet. The news item includes the suicide story of a high-school girl Cansel K. (17), who is claimed to have committed suicide after being sexually assaulted by her Maths teacher at the high school. Cansel's friends from the school claimed that her suicide was driven by the teacher's sexual assault, with whom the student was involved in an emotional affair. The school management was allegedly indifferent to this event although they were informed about the case. The teacher was arrested after the testimony of the witnesses.
[Honestly, you can't say you're going to school to get educated in a miniskirt like that. You could see this was coming. I wouldn't blame the teacher.]
In the extract given above, F.Ö. raises a standpoint based on the news he has read. The standpoint he defends can be reconstructed as "By wearing miniskirts at school, female students seduce their male teachers; therefore, when they get raped, the rapist should not be blamed". This standpoint faces two ad hominem arguments that suggest F.Ö. has the characteristics of a rapist. The commenter Ç.Ç. uses an argument which suggests that in order to hold such a standpoint, F.Ö. must/might have raped women with miniskirts or similar outfits. In parallel with Ç.Ç.'s comment, H.Ö. attacks F.Ö. by arguing that he looks like a rapist. In this way, H.Ö. makes a connection between the appearance of F.Ö. and the standpoint he expressed, implying that he defends this standpoint due to the fact that he must be a rapist himself. As a response to the abusive ad hominem attacks of Ç.Ç. and H.Ö., F.Ö. replies with an abusive ad hominem too, suggesting that the ones (either H.Ö. or F.Ö., or both) who attribute him the rapist quality are rapist themselves. The discussion comes to halt and no more comments were posted by the relevant people in the commenting area.

Refusing to Carry on the Discussion
In the second pattern of response to an abusive ad hominem attack, the commenter refuses to carry on the discussion. The pattern consists in such an exchange: a commenter, A, (who left a first-order comment concerning the news) faces an abusive ad hominem attack from another commenter, B, as a reply. The difference of opinion can be mixed or non-mixed. Later, A comes into the discussion to say that he refuses to reply to the comment of B (which includes the personal attack). A states this overtly and leaves the discussion 8 . Such a statement of refusal to contribute to the discussion is sometimes accompanied by a personal attack in return to B who has directed a personal attack against A before. This move indicates that A disqualifies B from being a serious discussion partner. 3 of the total 20 cases of responses to ad hominem attacks display this pattern of response. The dialectical profile for this pattern of response can be depicted as follows: A: +/p B: -/p because A is so-and-so ?/(+/p) because A is so-and-so A: maintain +/p and leave the discussion maintain +/p and leave the discussion because B is not worth because B is not worth responding to responding to The following case exemplifies the second pattern of response to an ad hominem attack: On March 17, 2016 a news item titled "The terrifying reason behind the deadly balcony escape" was posted to the Facebook page of the newspaper Hürriyet. The news item mentions a young lady, named Büşra, who tried to escape from the violence of her fiancé by climbing into a neighbour's balcony. The lady was able to get help from the neighbour and was taken to the hospital. She told the journalists that the whole evening she was tortured by her fiancé, and she had no option other than escaping through the balcony.
(2) S. The argumentation structures of the participants of the exchange in (2) can be reconstructed as follows: i) The argumentation structure of S.K.: (+/p) (Büşra deserves violence) ((+/p).1) (Her fiancé must have become a psycho due to her bad characteristics.) ((+/p).1).1 She must have made a cuckold of him.
iii) The argumentation structure of G.Ç.: (-/p) (Büşra does not deserve violence.) ((+/p).1) (S.K.'s contrary argument cannot be accepted.) (((+/p).1).1) (S.K. is a person of bad character.) ((((-/p).1).1).1) (She is worthless.) (((((-/p).1).1).1).1) (She is not even worth being spitted on the face.) In extract (2), S.K.'s standpoint can be reconstructed as "if a woman makes her fiancé go mad because of her bad characteristics, then she deserves violence". She provides a number of hypothetical arguments based on the appearance of the victim of violence such as having a fake look, cheating on her fiancé, and lying. The commenters M.K. and G.Ç. reply with abusive ad hominem. The personal attack by M.K involves attributing the characteristics of a slave to S.K. whose view is shaped by the dominating patriarchal thinking. G.Ç., on the other hand, argues that S.K. is worthless due to her opinion. S.K. comes back to the discussion later just to state that she refuses to respond to the claims of the commenters attacking her personally. Her response is accompanied by another ad hominem, calling the critical commenters such as M.K. and G.Ç. "idiots" and thereby implying that they are not worth responding to. After this comment, S.K. leaves the discussion.

Critically Evaluating the Abusive Ad Hominem Attack
In the third pattern of response to an ad hominem attack, the commenter critically evaluates the abusive ad hominem move and protests it as unreasonable. 3 of the total 20 cases of responses to ad hominem attacks display this pattern of response. The exchange can be represented as follows: Commenter A expresses a positive standpoint with regard to the proposition underlying the news item. Commenter B replies to this comment with an abusive ad hominem. The difference of opinion can be mixed or non-mixed. Later, Commenter A critically evaluates the ad hominem as an unreasonable argumentative move. This critical evaluation can be in the form of reason-giving why it is an unacceptable move in the discussion. One strategy that A can employ at this stage is to attack B's abusive ad hominem with a tu quoque argument suggesting that there is an inconsistency between what B says and what he does. Below is the dialectical profile depicting this pattern of response to an abusive ad hominem attack: A: +/p B: -/p because A is so-and-so ?/(+/p) because A is so-and-so A: maintain +/p because the maintain +/p because the argument that "A is so-and-so" argument that "A is so-and-so" is unreasonable.
is unreasonable.
The following case exemplifies the third pattern of response given to an abusive ad hominem attack: On March 6, 2018, a news item was posted to the Facebook page of Milliyet topicalizing a woman who was murdered by her lover. Muradiye B., the murdered woman, long time suffered from male violence. She got divorced from her husband due to domestic violence. Temporary debarment was imposed on the ex-husband by the court. Meanwhile, Muradiye, B. had an affair with Bülent Ö., with whom she worked at the same office. Finally, when she decided to get together again with the ex-husband, Bülent Ö. murdered her at their workplace and committed suicide afterwards.
In extract (3), Y.Ş.'s standpoint can be reconstructed as "Muradiye B. and her lover deserved death due to the fact that they were in a sinful affair and that Muradiye B. had a bad character". This standpoint is attacked by B.S. who puts forward that Y.Ş. does not qualify to be "human" as he gets pleasure out of the death of Muradiye B. B.S. argues that before one can judge women, they should be "human" first. Such a statement serves to prevent Y.Ş. from advancing his standpoint at the confrontation stage of the discussion and rule him out as a serious discussion partner. Facing the abusive ad hominem, Y.Ş. comes back to the discussion to evaluate B.S.'s personal attack critically. To this end, Y.Ş. uses a tu quoque ('you do it too') argument to state that the claims and practices of B.S. are at odds with each other. As for B.S., Y.Ş. is not fulfilling the qualities to be a respectful human, either, because of restricting Y.Ş.'s freedom to comment on the news he has read and calling him "ulan", which is a rude and derogatory address form in Turkish.

Discussion and Conclusion
In this study, i adopted the pragma-dialectical definition of an ad hominem attack as a fallacious argumentative move, committed by a party during the confrontation stage of a critical discussion, aimed at discrediting the opponent as a serious discussion partner by attacking the personal characteristics of the opponent rather than answering his/her arguments. The attack is a rule-violation in pragma-dialectical terms, namely the violation of the first rule of a critical discussion that assigns the discussants the freedom to advance and cast doubt on standpoints. In order to empirically observe how the discussion evolves after the personal attack and what effect an ad hominem move has on the discussion, I examined 20 responses to ad hominem attacks in online newscommenting in Facebook. The exchanges were organized in the following way: 'a first-order comment to a news iteman abusive ad hominem attack by another commenter against the initial commenter -the initial commenter's response to the abusive ad hominem attack.
The result of the analyses showed that an abusive ad hominem attack did create an obstacle to continuing the discussion in a reasonable way. Three patterns of responses were identified: (1) abusive ad hominem as a response to an abusive ad hominem attack; (2) refusing to carry on the discussion; and (3) critically evaluating the abusive ad hominem attack. All three patterns suggest that abusive ad hominem is regarded by the party who receives the attack as an unreasonable move that blocks the way to the resolution of the dispute. The abusive ad hominem attacks also motivate the attacked party to take measures to protect his image in the discussion. However, the discussion cannot be converted into a resolution-oriented one.
In 14 of the total 20 cases observed, a commenter responds to an abusive ad hominem attack with another abusive ad hominem attack (the first pattern). Such a repetitive pattern can perhaps be better understood by taking into account the sociolinguistic notion of 'face'. Neagu (2013) notes that abusive ad hominem attacks are taken as face-threatening acts undermining someone's positive face. Brown and Levinson (1987) describe "positive face" (1987) as a person's desire to be liked, appreciated, and approved. Once a person's positive face is attacked with an abusive ad hominem, that person has a tendency to return to the opponent with another ad hominem, which brings the discussion to a halt or makes it an infertile exchange of mutual attacks. Face-threatening acts are among the concerns of politeness theory; thus, future studies in this connection can further develop our understanding of ad hominem attacks.
3 cases displayed the second pattern of responses in which a commenter states explicitly that he/she will not respond to the ad hominem attack/s of the opponent/s. In this way, the commenter gives a clear statement that he will leave the discussion. Such a clear case of refusing to carry on the discussion is not common because, in most cases, the party who faces an abusive ad hominem attack does not come back to the discussion to give a reply. To put it differently, this commenter usually leaves the discussion silently. As the concern of this paper was to look at the explicit responses to abusive ad hominem attacks, such silent cases of leaving the discussion were not taken into consideration. Further studies may reveal that perhaps this is the most frequently chosen (silent) response to an abusive ad hominem attack.
Finally, 3 cases exhibited the third pattern of responses to an abusive ad hominem attack, namely, critically evaluating the abusive ad hominem attack. The critical evaluation of the abusive ad hominem by the attacked-party (the first-order commenter) involved showing by further arguments why this move was an unreasonable argumentative move in the discussion. One strategy that was used by the first-order commenter was to pay the party who committed the personal attack in his own coin. That is to say, the party who faced abusive ad hominem questioned whether the opponent who attributed negative characteristics to him was free of these negative traits himself. Once such negative characteristics were spotted, the first-order commenter could get back to his opponent with a tu quoque argument, stressing that the claims and practices of the opponent were at odds with each other.
A noteable feature of most examples was that the very first comment (by the first-order commenter) already includes an ad hominem fallacy; it is one that is directed to the persons involved in the news item (e.g. in extract (2) above, S.K's attack against Büşra, the woman involved in the news item). These people are not participants of the discussion; therefore, one might consider their role in the discussion only secondary. However, seeing that the persons involved in the news items are subjected to abusive personal attacks by some commenters, other commenters who emphatize with or ideologically associate themselves with those persons, attack the initial commenters with ad hominem moves (e.g. M.K.'s ad hominem attack against S.K. in an attempt to defend Büşra). The question may arise whether two wrongs (responding to a fallacy with another fallacy) can make things right. The analyses presented in this paper suggest that they cannot.
To conclude, in all three patterns of responses to ad hominem attacks that this paper illustrated, the discussion gets broken in the confrontation stage, the stage where the difference of opinion between parties becomes evident. Ad hominem fallacy is usually treated within a theoretical framework, and most examples that theoreticians use in explaining the fallacy are 'created' or 'unnatural' examples. This study is significant in that it involved empirical observation of the ad hominem fallacy in a real-life conversation context like Facebook news-commenting and monitor its effect on the whole conversation. Based on the observations of this study, we may conclude that the pragma-dialectical conception of ad hominem fallacy proves to be a functional definition of the fallacy considering that the responses given to ad hominem attacks show the discussion can no longer evolve in a constructive way. In this respect, such a discussion format barely contributes to deliberation aimed at opinion-formation concerning controversial social news.