A Comparative Analysis of the Use of the Discourse Marker “But” in a British Televised Political Interview Show: A Socio-Pragmatic Perspective

3.1 Corpora of the Study

This study employs a self-constructed corpus derived from the British political interview program HARDtalk, broadcast on the BBC. The rationale for choosing this program lies in its reputation as a leading and representative televised political interview show known for its rigorous and confrontational format. Its international acclaim for conducting in-depth and challenging political interviews makes it a prime candidate for the study. This setting naturally encourages the frequent use of discourse markers (DMs) like “but”, as interviewees (IEs) navigate tough questions and defend their positions. With its global reach and diverse array of IEs – including politicians and public figures from various cultural and linguistic backgrounds – HARDtalk provides a rich and varied corpus ideal for socio-pragmatic analysis in high-stakes political discourse, further enhancing the relevance of the findings. Specifically, the dataset comprises 60 one-on-one political interview episodes spanning the period from 2020 to 2023, yielding a total of 229,762 tokens. Within this dataset, the interviewer’s dialogue constitutes 40 % of the content, while the IE’s contributions represent 60 %, totalling 137,399 tokens, termed as The British Political Interview Corpus (BPIC).

Six sub-corpora have been created based on the IEs’ demographic attributes, namely, culture, first language, and gender, to facilitate the investigation. This approach is grounded in the recognition that the usage of DMs such as “but” can vary significantly across cultural, linguistic, and gender contexts, as highlighted in previous studies. When choosing data for analysis, close attention is paid to ensuring a balanced representation of the cultural backgrounds of the IEs involved in the program. This emphasis is placed because cultural background provides a more manageable and informative dimension, encapsulating a wide array of observable factors that collectively impact speakers’ communication styles, particularly in their use of discourse markers (DMs).

The definitions of culture are extensive and often debated (Spencer-Oatey and Kádár 2020). Regarding cultural categorization, IEs are classified into eastern or western cultural groups based on delineation of cultural values by Hofstede (2015). As suggested by Hofstede, Hofstede, and Minkov (2010), simplifying the complexity may involve expecting that the dimensions of national cultures reflect similar cultural differences. Recognizing the utility of generalizing about eastern and western cultures can be useful to highlight trends, this approach may overlook the intricacies and richness of global cultural diversity. For clarity, this study conventionally categorizes countries with shared cultural values as either eastern or western culture. For instance, the eastern group comprises IEs from China, HKSAR, Russia, India, and Singapore, whereas the western cohort encompasses primarily individuals from the United Kingdom, the United States, Sweden, Canada, and France. Consequently, the BPIC-E corpus consists of 30 IEs from the eastern culture, while the BPIC-W corpus comprises 30 IEs from the western culture. With respect to IEs’ first language, adherence to the criteria suggested by Müller (2005) leads to their classification as either native English speakers (NSs) or non-native English speakers (NNSs). Accordingly, the BPIC-NSs corpus comprises 26 native English speakers, while the BPIC-NNSs corpus consists of 24 non-native English speakers. In terms of gender distribution, the BPIC-F corpus comprises 21 female IEs, and the BPIC-M corpus includes 39 male IEs. It is noted that for the purposes of this study, these corpora exclusively feature discourses produced by IEs. To mitigate the potential impact of uneven distribution of IEs across these corpora, such as between BPIC-NNSs and BPIC-NSs, and between BPIC-F and BPIC-M, the study employs normalized frequency (N.F.) measurements. The distribution of the dataset is detailed in Table 1.

3.2 Methods

The investigation employs both quantitative and qualitative analyses. Firstly, the KWIC (Key-Word-in-Context) tool by Brezina, Weill-Tessier and McEnery (2021) in LancsBox is used to calculate the general distribution of the discourse marker (DM) “but” based on Crible (2018) definition. “But” is not regarded as a DM when it functions as a pure conjunction, a preposition signifying the meaning of “except”, or an adverb indicating “only”. Moreover, instances where “but” is presented in plural nouns (e.g., “ifs and buts”) and fixed expressions (e.g., “but for”, “but then”, and “cannot but”) are also excluded. The raw frequency (R.F) of the DM “but” in the six sub-corpora is computed, and the frequency ranking is then compared across the three demographic cohorts after the normalized frequency is calculated within each group using a basis of 10,000 words for normalization (Brezina 2018). Comparative analyses within each cohort are conducted using the Log-likelihood (LL) statistic by Rayson, Berridge, and Francis (2004), with statistical significance determined at thresholds of 3.84 for p < 0.05 and 15.13 for p < 0.0001. Critical values for statistical significance in the LL test are provided in Table 2.

Secondly, the functional analysis is executed on the generated concordance lines containing the identified DM “but”, which is indexed with a number and file name within LancsBox. This analysis utilizes functional model proposed by Crible (2018), drawing from taxonomies developed by Redeker (1990) and González (2004), which has undergone rigorous testing on corpus data, characterized by a high inter-rater ratio, adequately addressing all conceivable DM functions. This framework delineates 30 functions categorized across ideational, rhetorical, sequential, and interpersonal domains. For instance, the “ideational” domain encompasses eight functions elucidating objective relations among real-world events, while the “rhetorical” domain includes nine functions illustrating subjective relations between epistemic and speech-act events. The “sequential” domain portrays discourse structure through eight functions, while the “interpersonal” domain encompasses five functions focusing on the interactive management of the hearer-speaker relationship. In this study, the collective classification of “topic-shifting” and “topic resuming” functions within the “sequential domain” is designated as the as the “topic” function, as explicated in Table 3 of the DM functional taxonomy. In the process of annotating the functions, intra-rater reliability is assessed by calculating the percentage of raw agreement, a longstanding and valuable reliability measure (Brezina 2018). The author independently annotates the functions in two rounds, with a two-week interval between them, after becoming familiar with the functional scheme. The overall agreement is considered sufficient, with a value of approximately 94 %.

To evaluate the correlation between the functions of the DM “but” as employed by IEs and their demographic attributes, including cultural background, first language, and gender, statistical analyses such as the t-test and Mann-Whitney U test (Larson-Hall 2016) are applied. The selection of these tests is contingent upon considerations such as sample size and data characteristics, distinguishing between parametric and non-parametric criteria. For example, the t-test is employed to assess the correlation between IEs’ use of functions and their cultural groups (e.g., BPIC-E and BPIC-W), as well as to compare mean scores between the two groups. Conversely, the Mann-Whitney U test is utilized to investigate the correlation between IEs’ use of functions and their first language (i.e., between BPIC-NNSs and BPIC-NSs) and gender (i.e., between BPIC-F and BPIC-M). These tests serve as robust tools for scrutinizing the intricate relationships between the DM “but” and the demographic characteristics of the IEs in the study.

4.1 Frequency Analysis of the DM “but”

The BPIC corpus analysis reveals 732 instances of the discourse marker (DM) “but” employed by interviewees (IEs). Frequency distribution of this DM across six sub-corpora is depicted in Table 4. Specifically, IEs from the eastern cultural group utilize this marker 328 times, whereas their counterparts from the western culture employ it 404 times. The normalized frequency (N.F.) of the DM “but” in BPIC-W (N.F. = 54.1) is slightly higher than in BPIC-E (N.F. = 52.3). This indicates that IEs from the United Kingdom, the United States, Sweden, Canada, and France use the marker slightly more frequently than those from China, HKSAR, Russia, India, and Singapore. Despite the observed higher frequency in the western cultural cohort, the likelihood ratio (LL) test reveals no statistically significant disparity, as detailed in Table 5. Consequently, IEs’ cultural background does not appear to be a decisive factor influencing their utilization of the DM “but” in political interviews.

In terms of IEs’ first language, 380 instances of the DM “but” are identified in BPIC-NNSs, and 352 occurrences in BPIC-NSs, respectively. Native English-speaking IEs from the latter group (N.F. = 53.4) use the DM “but” at a rate almost equivalent to their non-native English-speaking counterparts (N.F. = 53.2), as depicted in Table 4. Statistical testing reveals no significant difference, suggesting that IEs’ first language may not play a substantial role in influencing their use of the DM “but” in political interviews. Concerning IEs’ gender, BPIC-F records 251 occurrences of “but”, while BPIC-M exhibits 481 cases. A marginal difference in normalized frequency, with a gap of 0.5 between the two groups, is observed. Specifically, female IEs display a normalized frequency of 53, while male IEs exhibit 53.5. However, statistical analysis indicates that this disparity lacks significance, thus suggesting that IEs’ gender may not significantly influence their utilization of the DM “but” during political interviews.

4.2 Functional Analysis of the DM “but”

An analysis of the utilization of the discourse marker (DM) “but” by interviewees (IEs) from the eastern and western cultural backgrounds uncovers distinct functional patterns. Within the eastern cultural group, three primary functions are observed, namely “contrast” (see Example [3] in Appendix 1), “concession” (see Example [4] in Appendix 1), and “cause” (see Example [5] in Appendix 1), predominantly manifesting within the ideational domain. Conversely, the western cultural group exhibits a broader spectrum, with nine identified functions spanning across four domains, including “contrast”, “topic” (see Example [2] in Appendix 1), “addition” (see Example [1] in Appendix 1), “concession”, “cause”, “reformulation” (see Example [8] in Appendix 1), “specification” (see Example [7] in Appendix 1), “approximation” (see Example [9] in Appendix 1), and “monitoring” (see Example [10] in Appendix 1). Detailed distribution of these functions and domains is presented in Table 6. An overview of examples for each identified function is found in Appendix 1. Notably, the predominant functional categories in both corpora are the “contrast” function and the “ideational” domain. Specifically, the instances of the “contrast” function constitute a significant proportion, with 95.1 % (312 cases) and 61.1 % (247 cases) in BPIC-E and BPIC-W, respectively, while the “ideational” domain encompasses 100 % (328 instances) and 66.6 % (269 instances) in BPIC-E and BPIC-W, respectively. This suggests a trend among IEs from both cultural groups to employ the DM “but” for contrastive purposes and conveying ‘factual’ information during political interviews. However, notable discrepancies exist in the frequency distribution of these functional categories between the two corpora. Particularly striking is the variance in domain distribution, where BPIC-W exhibits a dispersion of the identified functions across sequential, ideational, rhetorical, and interpersonal domains, whereas BPIC-E confines all three functions to the ideational domain, implying a more diversified usage among IEs from the western culture. Statistical analyses reveal a significant difference in both the types of functions (t(58) = 8.52, p < 0.001), and types of domains (t(58) = 13.4, p < 0.001). A higher prevalence of functional types is found in BPIC-W (M = 3.21, SD = 1.1) than in BPIC-E (M = 2.23, SD = 0.5). Similarly, the types of domains are also higher in BPIC-W (M = 1.3, SD = 0.5) compared to BPIC-E (M = 1, SD = 0).

Considering the first language of IEs, BPIC-NNSs and BPIC-NSs reveal distinctions in functional diversity and domain distribution. While six functions across three domains are observed in BPIC-NNSs, BPIC-NSs exhibit a broader range, with 10 functions across four domains. Notably, six overlapping functions, including “contrast”, “topic”, “addition”, “concession”, “specification”, and “cause”, are shared between the two corpora. However, BPIC-NSs exclusively feature four functions, namely “reformulation”, “approximation”, “monitoring”, and “condition” (see Example [6] in Appendix 1), along with the “interpersonal” domain, indicating a more nuanced usage pattern among native-English speaking IEs. The examination of the frequency distribution unveils a persistent pattern concerning the predominant functions and domains, wherein the “contrast” function constitutes 92.9 % (353 cases) and 60.2 % (212 instances) in BPIC-NSSs and BPIC-NSs, respectively. Likewise, the “ideational” domain is prevalent, representing 96.8 % (368 cases) and 65.3 % (230 cases) in BPIC-NSSs and BPIC-NSs, respectively. These findings underscore the pivotal role of the DM “but,” employed by interviewees (IEs), in conveying both contrasting and ‘factual’ information within the context of political discourse. Nevertheless, substantial disparities in the frequency distribution of these functions exist between the two groups, as evident in Table 6. Native-English speaking IEs seem to utilize a broader spectrum of functions and domains, a pattern corroborated by statistical tests indicating a significant difference in the types of functions (t(58) = 6.97, p < 0.001) and domains (t(58) = 8.82, p < 0.001). This indicates a greater usage of functional types in BPIC-NSs (M = 3, SD = 1.21) compared to BPIC-NSSs (M = 1, SD = 0.75). Similarly, the number of domain types is higher in BPIC-NSs (M = 2, SD = 0.51) than in BPIC-NSSs (M = 1, SD = 0.4).

In terms of gender groups, female IEs in BPIC-F employ eight functions across three domains, while male IEs in BPIC-M use eight functions across four domains. Despite slight variations in the frequency distribution of these functions between the two corpora, the “contrast” function and “ideational” domain persist as the top functional category in both BPIC-F and BPIC-M. Specifically, the “contrast” function accounts for 73.3 % (184 cases) in BPIC-F and 78.6 % (378 cases) in BPIC-M, while the ideational domain encompasses 78.9 % (198 instances) in BPIC-F and 83.2 % (400 instances) in BPIC-M. This indicates that both female and male IEs tend to employ the DM “but” for contrastive purposes and conveying ‘factual’ information in political interviews. A statistically significant difference between the two corpora is detected for the “topic” function (U = 21, p = 0.014), revealing a higher usage by female IEs (M = 4.43) compared to male IEs (M = 2.25).

5.1 Frequency

Despite minor distinctions observed among cohorts within the six sub-corpora, statistical analysis reveals no significant discrepancy in the frequency of “but” among IEs from three distinct demographic groups. This suggests that, from a statistical standpoint, IEs on HARDtalk exhibit consistent usage of the DM “but”, irrespective of cultural background, first language, or gender. IEs from both western (e.g., the United Kingdom, the United States, Sweden, Canada, and France) and eastern cultures (e.g., China, HKSAR, Russia, India, and Singapore) demonstrate comparable frequencies in employing “but”, contrary to previous claims about cultural impacts on discourse markers (DMs) (Chen and Wang 2021; Hofstede 2015; Huo 2023; Zand-Moghadam and Bikineh 2015), which might be attributed to the characteristics of political interviews (Clayman and Heritage 2002), where linguistic convergence occurs due to the need to assert and defend positions. Both native and non-native English-speaking IEs show similar patterns in using “but”, indicating non-native speakers’ competence in political discourse (Buysse 2020; Deng 2021; Huo 2023; Zaykovskaya 2022). In the confrontational setting of political interviews, DMs like “but” help signal contrast or opposition, essential for clear and assertive communication. Gender also does not significantly affect the frequency of “but” usage, with both female and male IEs displaying similar rates. This finding opposes earlier research on gender-specific DM usage highlighted by Ebrahimi and Xodabande (2023), Virginia Acuña (2021), and Tavakoli and Karimnia (2017), as well as Kozubíková Sandová (2014), highlighting the universality and neutrality of usage patterns in political interviews on HARDtalk. This lack of significant differences in “but” frequency among demographic groups underscores the complexity of language use in political interviews. Various factors, such as sample size limitations, within-group variability, and interaction effects, may influence this complexity (Altıparmak 2022; Deng 2021; Torgersen, Gabrielatos, and Hoffmann 2017; Zhang et al. 2023). Further research with larger, diverse samples and consideration of contextual factors is needed to better understand DM usage in political communication (Chilton 2004).

5.2 Function

The functional analysis conducted in this study illustrates the multifaceted nature of the DM “but”, echoing the polyfunctional nature typical of discourse markers (Ranger 2018). This observation demonstrates the varied roles that the DM “but” assumes within political interviews, irrespective of IEs’ socio-pragmatic factors, as evidenced by its involvement in various functions such as “contrast” (see Example [3] in Appendix 1), “topic” (see Example [2] in Appendix 1), “addition” (see Example [1] in Appendix 1), “concession”, and “cause”. In Example [3] (see Appendix 1), the use of “but” serves to draw a clear distinction between the actions or characteristics of the two groups. While the first part of the statement suggests a similarity or continuity between the cyber army and later groups like Anonymous, the “but” introduces a significant difference, specifying that the level or type of activity, particularly actions involving the compromise of computers, was less pronounced in the cyber army group. A frequency analysis of these functions reveals a consistent and predominant usage of the DM “but” among IEs from all three cohorts for conveying contrasting opinions, thus reflecting its asserted core meaning (Fraser 2009; Schiffrin 1987). The prevalent use of the “contrast” function aligns with the communicative goals of IEs and the confrontational and adversarial characteristics of political interviews (Clayman and Heritage 2002). Such interviews, exemplified by programs like HARDtalk (Mullany 2002), foster dynamic exchanges characterized by probing inquiries and robust debate. The deliberate insertion of contrasting ideas through the use of “but” allows IEs not only to strengthen their arguments but also maintain control and influence over the discourse.

In examining the functional categories, the “ideational” domain emerges as the most frequently employed by IEs from all three groups, regardless of their socio-pragmatic features. This consistent utilization of the ideational domain suggests IEs’ prioritization of conveying ‘factual’ information during their interactions on HARDtalk, reflecting a commitment to presenting accurate data to substantiate their claims and engage in substantive discussions. This highlights the importance of evidence-based reasoning in political discourse and echoes the neutrality of political interviews (Hutchby 2022). According to Clayman and Heritage (2002), during the process of responding to the interviewer’s inquiries, the IE consistently refers to “the same third party” to support previously stated perspectives, thereby affirming the interviewer’s “naturalistic stance” (p. 163). Additionally, by grounding their statements in verifiable facts and data, IEs establish themselves as knowledgeable and credible sources, projecting professionalism and authority. This strategic use of the ideational domain may also shape audience perceptions of IEs’ competence, trustworthiness, and persuasiveness, thereby enhancing their effectiveness in conveying messages and gaining public support for their positions (Johansson 2007). In Example [4] (see Appendix 1), the interviewee acknowledges a significant amount of money collected during their time in office. This acknowledgment sets up an expectation that the money might belong to the government. However, the use of “but” introduces a contrasting fact: the money does not belong to any government, including the current Taliban government. Instead, the money belongs to the Afghan people. The reasoning presented by the interviewee is grounded in factual information rather than subjective opinions: The money was collected during their tenure; The money does not belong to any government (past, present, or future); The money belongs to the Afghan people. By providing these factual assertions, the interviewee employs “but” to logically contrast these facts, minimizing subjective input and focusing on a clear, objective concession. IEs use “but” less frequently in rhetorical and interpersonal domains due to the formal, fact-oriented, and structured nature of these interviews. They prioritize presenting objective information, defending their viewpoints, and adhering to the interview’s agenda. The use of “but” primarily occurs in the objective and sequential domains to introduce contrasting information and manage the logical progression of the discussion, aligning with the expectations of objectivity, professionalism, and clarity that characterize political interviews on platforms like HARDtalk.

The examination of predominant functional categories through comparative analysis does not yield statistically significant differences among IEs within each cohort, suggesting a degree of universality in communicative functions and linguistic behaviours associated with the utilization of “but” across IEs. This consistent pattern implies a convergence in communicative strategies and a shared comprehension and application of “but” as a versatile tool for indicating contrast and conveying ‘factual’ information across diverse demographic cohorts. This observation resonates with the assertion by Fraser (2009) regarding the universality of basic DMs such as “but”. From a socio-pragmatic perspective, it challenges conventional assumptions about the impact of demographic factors on DM usage and underscores the intricate interplay between these factors, communicative objectives, and contextual influences in shaping the usage patterns of DMs like “but” during political interviews, contributing to a deeper understanding of the universality and adaptability of linguistic behaviours within the domain of political discourse on televised platforms like HARDtalk.

Regarding cultural aspects, the identification of IEs from the western culture as employing a wider array of functions of the DM “but” compared to their counterparts, with statistically significant differences, offers intriguing insights into the interplay between cultural background and linguistic behaviour within political discourse. This disparity may be attributed to the western communicative norms that prioritize assertiveness, individualism, and directness (Barbara, Afzaal, and Aldayel 2024; Cheng et al. 2021). Consequently, Western IEs may use “but” for various purposes, such as signalling contrast, introducing new topics, expressing concession, or highlighting emphasis, reflecting their strategic adaptation to the confrontational nature of political interviews. Statistically, native English-speaking IEs exhibit a broader range of functions compared to non-native counterparts, likely due to higher linguistic proficiency and familiarity with English. Native speakers’ deeper understanding of English DMs enables nuanced use of this marker, while non-native speakers may prefer straightforward usage for simplicity and clarity, aligning with studies conducted by Fung and Carter (2007), Liu (2017) and Huo (2023).The analysis also shows that female IEs use the “topic” function more often than males, possibly due to gendered communication styles (Virginia Acuña 2021). Women tend to favour cooperative and inclusive communication, emphasizing rapport-building. In political interviews, female IEs may use the “topic” function to foster dialogue, acknowledge alternative viewpoints, and shift the conversation focus, aiming to broaden discussion scope and cultivate inclusive discourse. The female interviewee in Example [2], utilizes “but” to shift the topic and refocus the conversation on a crucial point regarding the systemic issue within USA Gymnastics. The interviewee begins by discussing general perceptions of mental health, trauma, and the challenges of speaking up about abuse. She then employs “but” to introduce a new, a more specific topic (the presence of a pedophile within USA Gymnastics) that is directly linked to the broader discussion of trauma and abuse but shifts the focus to a particular instance of systemic failure. This shift serves to ground her general statements in a concrete example, emphasizing the gravity and historical persistence of the issue, as well as highlighting the intersection of her identity and the narrative, suggesting a deeper, personal understanding of the subject matter. Through this, the interviewee also underscores the urgency and seriousness of the issue, which might be more resonant coming from a female perspective, particularly in the context of a male-dominated system (both in gymnastics and broader societal structures).

In summary, these findings offer valuable insights into the nuanced ways in which IEs’ socio-pragmatic characteristics may shape their linguistic behaviours, particularly in the use of the DM “but” in the context of political interviews.