The Effect of Discourse Mechanisms of Break on the Shopping Attitude of Social Network Users in the Comment Marketing

Comment marketing is a digital strategy that uses hashtags and mentions to attract users to websites or Instagram pages. This study sheds new light on the effects of topic drift in comment marketing on customers' shopping attitudes. Accordingly, a mixture of qualitative and quantitative methods was used. A sample of 5,000 views from the most popular posts by Digikala Holding Company in 2021-2022 was randomly selected for a qualitative investigation. Following that, 460 views were purposefully selected to be investigated using the integrated research method. Following a review of the most relevant research related to the research variables and based on the qualitative results, the conceptual model of the article was developed. Then, based on the conceptual model of the research, a questionnaire was conducted. After approving the content validity and reliability of the questionnaire, it was administered to 366 Instagram users. Based on research data obtained through a survey and questionnaire test, the analysis of the structural equation method was used to create a pattern based on the study's statistical population. The results showed that Instagram users use various break mechanisms - parasite, mutual referential, and republishing - to disrupt the speech chain to attract users to their Instagram page or website. Furthermore, the break mechanisms of topic drift in the marketing strategy have a linear, positive, and significant effect on social network users' shopping attitudes. Compared to the parasite and republish mechanisms of the break in comment marketing, the mutual reference mechanism positively impacts users' shopping attitudes.





1. Introduction

The digital revolution has placed a whole new set of capabilities in the hands of consumers and businesses. Following that, marketing has not only focused on satisfying consumer needs. Instead, it prefers to empower and enable users to construct the realities which they intend to experience through the construction of communities. Due to this, digital marketing through social networking enables companies to communicate with their customers at relatively low costs. Thus, commenting is one of the online marketing methods that allow users to introduce their content to their website through the sharing of an access link toward their Instagram page, website, or other websites.

In comment marketing, companies often try to direct the customer to their page by changing the topic or breaking the chain of speech during the discussion of a brand post. In other words, by examining the conversation process - which can be short or long - one can see that the topic at the end of a conversation is very different from that of the beginning. This topic shift has occurred gradually and users are usually unaware of this event. In fact, the ideal conversation is not one in which the chain of turns revolves around a single topic. Instead, there should be a proper balance between the steps that are related to the topic and the steps that take the conversation in a new direction.

From the viewpoint of discourse as a process, communication and coherence are the consequences of the interaction between users that can be achieved by mutual efforts of both the communicator and the audience. Accordingly, this research is aimed at answering the question that "in the online written conversation, what mechanisms of topic drift are used in the comment marketing to affect the users' shopping attitudes?" Therefore, the first hypothesis of the study is: "The use of topic drift mechanisms of break in comment marketing can affect social network users' shopping attitudes" (The main hypothesis). According to this hypothesis, there are three sub-hypotheses: first, the use of parasitic mechanisms in the comment marketing strategy negatively affects the shopping attitude of users (sub-hypothesis one)." Secondly, the use of mutual reference mechanisms in comment marketing has a positive effect on the shopping attitude of users (sub-hypothesis two). Lastly, the use of republishing mechanisms in comment marketing has a negative impact on the shopping attitude of users (sub-hypothesis three).

In order to discover unknown layers of meaning in a systematic way, a mixed method consisting of qualitative and quantitative methods was applied. Following a review of the most relevant research related to the research variables and drawing upon the qualitative results, the conceptual model of the article was developed. Then, on the basis of a conceptual model of the research, a valid and reliable questionnaire was conducted. Based on research data obtained through a survey and questionnaire test, the analysis of the structural equation method was used. This analysis is used to create a pattern based on the statistical population of the study.



2. Literature Review

Due to the fact that communication in cyberspace has led to the emergence of extensive communication platforms among the customer (Baker, 2003; Kotler & Keller, 2006; Fırat & Dholakia, 2006; Kotler et al., 2018; Ajina, 2019; Algharabat et al., 2018; Kapoor et al., 2018; Kaur et al., 2018; Lal et al., 2020; Lopes & Casais, 2022), commentary has become a central issue in digital marketing (Bahtar & Muda, 2016; Aibing, 2018). By developing the strategy of comment marketing (Lee et al., 2020; Aibing, 2018;), companies often try to direct the customer to their page by changing the topic or breaking the chain of speech during the discussion of a brand post (Eisend, 2016; Lee et al., 2020).

However, by looking at the conversation process, one can see that the topic of the conversation ended up being very different from the topic at the beginning of the conversation (Razeghi et al., 2020a; Razeghi et al., 2020b; ). This topic shift, or according to Hobbs (1990, 1) topic drift in the written dialogue (Danesi, 2018; Nouruzi & Arjmandi, 2021; Yus, 2010) has occurred gradually and users are usually unaware of this event (Razeghi et al., 2020a; Tanskanen, 2006). In this regard, coherency (Halliday & Hasan, 1976; Mann & Thompson, 1987; van Dijk, 1977; Schank, 1977; Daneš, 1978; Goutsos, 1997; Halliday, 1985; Abbasi, 2001) and topic drift in discourse analysis (Sarnovsky & Kolarik, 2021; Button & Casey, 1984; Myers, 1998; Hobbs, 1990; Herring & Nix, 1997; Herring, 2003; Razeghi et al., 2020 b) which can inevitably be a controversial issue has not been considered. By considering the vital role of user shopping attitude in the field of marketing, various types of research from a psychological point of view (de Matos et al., 2007; Ajzen & Fishbein, 1980; Schiffman & Kanuk, 1997) in relation to marketing (Kotler et al, 2018; Setiyawati et al., 2017; Holt, 2002; Touzé, 2020; Eisend, 2016) have applied this concept. Meanwhile, the contribution of linguistic and discourse research in this field is very small and is limited to only a few studies (Phillips & Hardy, 1997; Park & Lee, 2019). Despite this interest, no one to the best of our knowledge has studied comment marketing from a discourse analysis point of view on shopping attitude.



3. Methodology

Digi-Kala Instagram's page as one of the most popular holding companies was analyzed during 2021-2022. Instagram had a lower filtering rate during this period when compared to other social networks in Iran, increasing its engagement rate. For this purpose, a mixture of qualitative and quantitative methods was used to reveal hidden layers of meaning in a systematic way. As part of a qualitative investigation, a sample of 5000 views published under the most viewed posts published by Digikala's Holding company in the period of 2021-2022 was randomly selected. Following that, 460 views were purposefully selected to be investigated using the integrated research method. Following a review of the most relevant research related to the research variables and based upon the qualitative results, the conceptual model of the article was developed. Then, based on the conceptual model of the research, a questionnaire was conducted. After approving the content validity and reliability of the questionnaire, it was administered to 366 Instagram users. Based on research data obtained through a survey and questionnaire test, the analysis of the structural equation method was used to create a pattern based on the statistical population of the study. After that, the software application used to analyze the data was Amous (0.89 version).



3. Theoretical basis

3. 1. Integrated model

Coherence and topic relevance is a macro-semantic structure, also called the subject of discourse (van Dijk, 1977). It is formed hierarchically by smaller structures (van Dijk, 1977). In conversations, there are general and specific topics that are either covered by the whole conversation or simply by one utterance. Moreover, topics are included in one another and organized in a kind of hierarchy of sub-topics inside topics. In stepwise movement, "any next utterance is built in such a way as to be on topic with a last" (Sacks, 1992b, 300). The stepwise movement involves the development of a topic, while the boundary movement concerns the boundary between topics. In the development of a topic, there is another type of movement which is less controlled and slower than stepwise movement which is "topic drift" (Hobbs, 1990).

In conversations, we find general and specific topics that cover the whole conversation or are contained within one utterance. Moreover, topics are included in one another and organized in a kind of hierarchy of sub-topics inside topics. In stepwise movement, "any next utterance is built in such a way as to be on topic with a last" (Sacks, 1992b, p. 300). In this sense, a stepwise movement concerns the development of a topic whereas a boundary movement concerns the limits between topics.

In the development of a topic, there is another type of movement which is less controlled and slower than stepwise movement which is "topic drift" (Hobbs, 1990). Hobbs (1985) classified topic drift as 1) a kind of association based on sematic parallelism, 2) chained explanations, or explanations that become topics in their own right, and c) metatalk (evaluation) used as a way of introducing a novel topic (Hobbes, 1985, 3-9). Hobbs (1985) suggests that subsequent discourse segments shift the subject through parallelism, explanations, or metatalk. Herring and Nix (1997) added break to this division. Parallel moves include the introduction of different entities with the same properties as those already mentioned, or other properties of the same entities. Explanations expand on the topic at hand by explaining a previous proposition. Metatalk serves to structure the discourse. Break changes the topic. (Herring and Nix, 1997, 4). In addition to this break, Razeghi et al (2020) add three subclasses: A) Parasite: advertising text intended to be followed by users, B) mutual reference, mention to other matters outside or other sides of the Instagram page, C) republishing: tagging users or mentioning a subject by using # or @ (Razeghi et al., 2020). In general, the mechanism of topic drift and break is shown in the below figure.



Figure 1

Coherence and Topic drift mechanisms

















Similarly, the soft and flexible nature of the market in social networks constantly encounters consumers with novel products to influence their shopping attitude (De Matos et al., 2007). Attitude refers to how people view the world. Consumers’ attitude is guided by their beliefs which determines their behavior and purchase intentions. In general, intentions to perform a behavior will be influenced by individual and interpersonal characteristics (Kotler and Armstrong, 2018).

A large and growing body of literature has investigated the coherence and the topic of discourse (Hobbs, 1990). In addition, more recent attention has focused on the effects of topic drift on conversation flow (Razeghi et al, 2020). Previous work has only focused on the strategy of comment marketing, but the mechanisms of topic drift in comment marketing have not been studied so far.



3. 2. Conceptual model

The qualitative findings in this study revealed that some users, by publishing their comments, attempt to break the written discourse chain. In addition, they direct traffic by linking to their own website or Instagram page. Hence, it could conceivably be hypothesized that "the usage of topic drift mechanisms in comment marketing has an impact on the shopping attitude of social network users" (The main hypothesis)". This hypothesis itself consists of three sub-hypotheses as "by using Parasite, mutual reference and republishing mechanisms in the comment part of a post, we can have a negative effect on users' shopping attitudes (Hypothesis one)."

In general, on the base of previous studies and our research qualitative investigation, the following conceptual model was obtained, which was used to measure the users of the questionnaire as follows:





Figure 2

Conceptual model of research











4. Results

Research data will be analyzed using structural equation modeling in this section. The structural equation method can study the causal relationships between variables. In general, the hypothesis under consideration in a model of structural equations is a causal structure observed among a set of invisible structures. These structures are measured by a set of observer variables. In this research, Amous software has been used to analyze the structural equations of the proposed research model (Ebrahimi et al., 2022).

Structural equation analysis can be explained with two models of measurement and structural. In this regard, by confirming the data fitness, the investigation of each model will be complete. Fitness is a scale between 1 and -1. Accordingly, as long as a variable's fitness amount is close to one, its effect is higher. Therefore, if the number 1 is obtained, it is the exact variable. If this value is -1, it means that the variable in question has an effect but its effect is reversed.



4. 1. Measurement model

The fitness of our conceptual model is reported based on a set of fitness indicators in the table below. From all the mentioned indicators, it can be seen that the research model has a good fit with the studied data.



Table 1

Model fit (linear regression)

Estimate S.E. C.R. P Label

Shopping attitude <--- Break .446 .065 6.908 ***

Q5 <--- Break 1.000

Q6 <--- Break .999 .049 20.588 ***

Q7 <--- Break .998 .047 21.291 ***

Q8 <--- Break 1.015 .049 20.623 ***

Q9 <--- Break .965 .049 19.772 ***

Q10 <--- Break .829 .052 15.991 ***

Q16 <--- Shopping attitude 1.000

Q15 <--- Shopping attitude .495 .162 3.061 .002

Q14 <--- Shopping attitude .328 .149 2.198 .028

Q13 <--- Shopping attitude .665 .169 3.929 ***

Q12 <--- Shopping attitude .952 .191 4.984 ***

Q11 <--- Shopping attitude 1.825 .279 6.530 ***



The table above shows that the relationship between the latent variables and the shopping attitude is significant.

Also, the table below shows the standardized regression coefficient of the relationships between the variables. In general, the standardized regression coefficient of the relationships between the variables, which is often 0.5, indicates the high intensity of this relationship. In general, based on the table below, it can be seen that the relationship between research variables is positive and direct.







Table 2

Model fitness (standardized linear regression)

Estimate

Shopping attitude <--- Break 1.000

Q5 <--- Break .835

Q6 <--- Break .859

Q7 <--- Break .877

Q8 <--- Break .860

Q9 <--- Break .838

Q10 <--- Break .727

Q16 <--- Shopping attitude .359

Q15 <--- Shopping attitude .180

Q14 <--- Shopping attitude .124

Q13 <--- Shopping attitude .246

Q12 <--- Shopping attitude .352

Q11 <--- Shopping attitude .660



According to the fitness model, there is a very high and acceptable relationship between the variables. By studying the questionnaire's internal structure and discovering the independent and latent variables contained within each structure, it was determined that the structures used to measure the independent and present variables are quite appropriate and have the ability to measure these variables.



4. 2. Structural model

The tested conceptual model is presented in the figure below. The numbers written on the lines are actually beta coefficients from the regression equation between the variables, which is the path coefficient.









Figure 2

Tested research model (path coefficients and operating loads)





The above structural model shows the internal relationship between research variables. Indices of the latent variable "topic drift" were evaluated with 12 independent variables. Among these indicators, break with 6 independent variables (parasite with 2 independent variables, mutual reference with 2 independent variables, and republishing rupture with 2 independent variables) were measured. Also, the indicators of shopping attitude as a latent variable were measured by 6 independent variables which were positive, linear, and meaningful.

In general, by analyzing the internal structure of the questionnaire and discovering the constituent factors of each construct or latent variable, and using confirmatory factor analysis tools, it was found that the structures used to measure latent variables are quite suitable for the ability to measure these variables.

In sum, the aim of this study was to investigate the effect of break mechanisms of topic drift on the shopping attitude of social network users. In this regard, some users attempted to break the chain of written discourse by publishing their comments in order to introduce other brands' Instagrams/websites. Based upon quantitative investigation, it was founded that Instagram users benefit from various break mechanisms including parasite, mutual referential, and republishcation to disrupt the speech chain and then attract users to their Instagram page or website. Furthermore, the break mechanisms of topic drift in comment marketing have a linear, positive, and significant effect on social network users' shopping attitudes. Compared to the parasite and republish mechanisms of the break in comment marketing, the mutual reference mechanism has a positive impact on users' shopping attitudes. In this regard, the optimal agreement between the structured model and the experimental data was studied. Then, by applying the structural equations, an appropriate model was designed for the relationship between latent and explicit variables. Finally, the research findings showed that interactive discourse mechanisms can play an influential role in comment marketing due to their capabilities. It is undeniable that identifying and discovering elements of the marketing discourse perspective can contribute to changing users' attitudes as well as building a lasting connection with them.