1. Introduction¹

During the last decade, there has been a growing need for researchers to adapt to new socioeconomic and cultural demands that reflect a shift in the creation, dissemination, and access to scientific knowledge. This is a consequence of Open Science policies advocating for a transparent and open sharing of research to expert and non-expert audiences (Luzón and Pérez-Llantada 2022). Some of the requirements that researchers face include gaining international visibility and recognition, meeting institutional standards, and securing public funding. To meet these evolving demands, researchers and scholars have embraced a range of digital genres that enable the dissemination of scientific knowledge to wide diversified audiences. Traditionally, ‘genre’ is understood as a communicative event with specific form conventions, targeting specific discourse communities, and fulfilling social actions (Miller 1984; Swales 1990). However, in the present digital landscape, researchers have access to digital resources and tools that enable them to share diverse data and findings with broader audiences, therefore relying on “new possibilities for interactivity and collaborative construction” of knowledge and participatory communication practices out of their discourse community (Belcher 2023: 38).

Relevant to digital genres in the context of Open Science is the notion of ‘transformative science’, which refers to the use of innovative online communication practices to disseminate scientific knowledge addressing the lay public and academic peers. Pérez-Llantada et al. (2022) reported that these transformative practices involve researchers using open-access repositories when sharing pre-prints and papers, academic social networks to stay updated on the latest developments in their respective fields, or social media platforms such as Twitter (X), Facebook, Reddit, and Instagram to communicate scientific research to special interest groups and broad audiences. It is well-known that among the various social media used by researchers, Twitter has gained prominence as a preferred medium for sharing scientific knowledge with both the lay public and scholarly peers due to its instant and short messaging nature. According to authors such as Büchi (2016), Lee et al. (2017), Côté and Darling (2018), Mehlenbacher (2019), and Tardy (2023), researchers join Twitter mainly to disseminate their work, promote their research outputs and publications, and network with colleagues in their disciplinary fields.

Research on Twitter usage by scholars has received attention because of its potential to make scientific knowledge accessible to diversified audiences (Darling et al. 2013; Lee et al. 2017; Luzón and Albero-Posac 2020). For instance, Darling et al. (2013) explored the usefulness of Twitter during the publication process as they analysed exchange practices among colleagues to generate ideas, receive peer-review comments, and increase the impact of their manuscripts’ contents. Similarly, Lee et al. (2017) and Luzón and Albero-Posac (2020) investigated the practices of networking and communication in specific academic scenarios, particularly in academic conferences, where Twitter has become a powerful tool that combines with the on-site conferences as a means for informal and formal communication. In this respect, Luzón and Albero-Posac (2020) identified four main communicative functions of conference tweets that have organisational purposes, promote informal interaction, foster community building, and focus on networking.

Further research on Twitter has focused on the rhetorical analysis of tweets written by scientists and public organisations (Orpin 2019; Tardy 2023), the use of digital affordances such as hyperlinking, multimodal composition, and intertextuality characteristic of academic tweets (Büchi 2016; Luzón 2023), or the different roles played by scientists when communicating science outside academia (Walter et al. 2019). Additionally, the analysis of the combination of different semiotic resources and Twitter affordances in tweet composition is a significant line of research in applied linguistics, where elements such as visuals, videos, hyperlinks, mentions, hashtags, and retweets, among others, are investigated to know how they help to disseminate messages and effectively engage wider audiences (Orpin 2019; Luzón and Albero-Posac 2020; Luzón 2023; Tardy 2023; Villares 2023a; Xu et al. 2023).

While previous research has tended to focus on multimodality and hypertextuality, the third key feature of digital genres ––interactivity–– has received less attention in the literature on Twitter for academic purposes. Interactivity on Twitter has examined the type of participants and readership of tweets (e.g., Walter et al. 2019), yet a deeper analysis from a discursive perspective has not been conducted yet. Tardy (2023) points out that scientific communication still occurs in its majority among audiences who are knowledgeable on the topics rather than reaching readership outside academia, so it seems relevant to examine how communicative exchanges between specialised audiences occur. To explore this issue, this paper analyses a corpus of academic tweets from an emerging digital genre called Twitter Conference Presentation (henceforth, TCP), which has remediated the traditional on-site academic conference presentation into the digital medium (Villares 2023a, 2023b). The TCP consists of a six-tweet thread where presenters share their research projects. Like on-site conference presentations, the TCP can be followed by a discussion session in the form of tweets that readers can post at the end of each thread to engage in a conversation with the presenter (and/or other readers). Based on the literature, it is hypothesised that even though the presenter cannot control who reads and responds to their content, they still want to initiate discussions and interact with potential readers. In order to give insights into how and why researchers may engage in online conversations that take place on academic Twitter, the present study analyses the discussions following TCPs to address the following research questions:

1. Do Twitter Conference Discussion Sessions (henceforth TCDSs) follow the same interactional turn-type patterns as traditional on-site academic conference discussion sessions?
2. What are the main communicative functions and purposes of tweets in TCDSs?
3. Does the medium shape the type of digital discourse features participants use in their TCDS tweets?

Section 2 is devoted to an overview of the on-site conference discussion session. After that, Section 3 delves into the corpus description, data collection process, and analytical techniques. Section 4 reports the results in terms of turn-types, communicative functions supported by rhetorical strategies, and an exploration of digital discourse affordances. Finally, the paper concludes with a discussion of the main findings and their implications for researchers’ communication skills development.

2. The conference discussion session

The emergence and constant evolution of digital genres sometimes bring changes in the form, functions, and communicative purposes of traditional academic genres. In the case of the TCP, it still shares the primary communicative goals and functions of face-to-face conference presentations, that is, presenting work in progress and networking (Rowley-Jolivet and Carter-Thomas 2005; Hyland 2009), while introducing novel discursive and rhetorical strategies that arise from the affordances and constraints that Twitter (e.g., hashtags, mentions, retweets, or space restrictions) and the digital medium offer (Tagg 2015; Zappavigna 2017).

The conference presentation is part of a genre chain that consists of a series of genres organised in a chronological sequence. The conference presentation is preceded by genres such as the call for papers and the abstract, while it is followed by the conference paper and the discussion session (Räisänen 2002). The discussion session is defined as “the event that takes place right after a presentation at an academic conference in the form of dialogues between the ‘presenter’ and the ‘discussants’” (Xu 2022a: 63). Regarding the organisation of interaction between participants in the discussion session, Querol-Julián and Fortanet-Gómez (2012, 2014) identified its rhetorical and turn-taking structure, shedding light on its distinctive nature, and highlighting the combination of linguistic and non-linguistic features during the turn-taking exchanges. The three types of turns identified by the authors are: a) comments ––when a turn includes a statement–– b) questions ––when a turn includes at least one question––, and c) responses from the presenters. In more detail, the generic structure of the dialogic exchange starts with the discussant’s question followed by the presenter’s turn (response). The question can include the following moves: a) announcing the question, b) asking the question, and c) reformulating the question. On the other hand, the presenter’s response may consist of a reaction to the question, answering the question, expanding the topic of the question, and closing the turn (Querol-Julián and Fortanet-Gómez 2014: 86).

Concerning the main communicative functions of turns, discussant turns are characterised by an evaluative function and specific lexico-grammatical features attached to that function. Drawing upon Webber’s (2002) comprehensive account of question types and participants’ reactions during discussion sessions, it is possible to classify question functions into five main categories: a) fact-seeking questions, b) opinion-seeking questions, c) justification-seeking questions, d) suggestions, and e) neutral statements. There is a gradual evaluative function in the different turns, ranging from low evaluative turns (e.g., facts or statements) to high evaluation when criticism appears (e.g., justification-seeking questions). In order to reduce the potential threat of criticism in a turn, Xu (2022a) argued that building rapport was used as a common practice among discussants who dedicated more effort to thank and praise the presenter at the beginning of their turn. Rhetorical strategies to soften criticism can take the form of hedges (e.g., I think), admission of limitations (e.g., I don’t know), and evaluative language to show appreciation (Webber 2002; Hyland 2005; Xu 2022b). Hence, rapport-building strategies that contextualise and introduce a comment rely on discursive features of politeness and solidarity, alongside other lexico-grammatical features, which tend to be employed when there is a high level of evaluation (Wulff et al. 2009; Xu 2022a). Additionally, Konzett (2012) ––in her book about identity construction at academic conferences–– noted that the purpose of raising a question may extend beyond seeking or evaluating scientific information, to encompass aspects of negotiating professional identities and self-promotion. In sum, a discussant’s turn may include different communicative purposes other than seeking and exchanging knowledge.

3. Methods and procedure

3.1. Corpus description

The corpus comprises 561 tweets (13,105 words) posted in the discussion sessions of 55 presentations in English² from the Twitter Conference Linguistweets (ABRALIN 2020).³ Yet, the corpus is multilingual and includes tweets in English (81%), French (10%) Portuguese (5%), and other languages (4%). There is a total of 235 discussions. On average, each presentation received ten tweets and up to three users interacted in the discussion. Examining the composition of the corpus in more detail, as shown in Table 1, tweets were organised into ‘comments’, ‘questions’, ‘responses’, ‘automatic comments’, and ‘follow-up conversations’ whenever a discussion included more than two tweets.

Turn-type		Number of tweets	Percent
Automatic comment		2	0%
Comment		158	28%
Question		89	16%
Response		169	30%
Follow-up conversation		143	25%
	Follow-up comment	77	14%
	Follow-up question	19	3%
	Follow-up response	47	8%
Total		561	100%

Table 1: Corpus description

Questions and comments are tweets written by readers that initiate the discussion, while responses are the presenters’ replies. Sometimes a response can consist of more than one tweet. The category follow-up conversation refers to a discussion that involves more than just the standard turn-taking sequence of comment-response or question-response. 25 per cent of the corpus tweets belonged to this category, which involved longer interactions, where participants engaged in longer exchanges repeating the comment-response pattern. Finally, a medium-related category was identified ––‘automatic comments’–– which consisted of tweets automatically generated by a software to promote presentations.

3.2. Corpus collection

Tweets were accessed at the conference website (https://www.linguistweets.org/linguistweets-2020/programa/), manually downloaded from the participants’ Twitter accounts, and stored in Word documents so that both text and other semiotic resources (e.g., images, emojis) could be analysed. 55 documents were compiled to store the TCDSs separately and filed under an anonymous name, e.g., TCDS1 for the tweets comprising all the discussions associated with the first presentation. For each TCDS, the following items were identified: title of the presentation and hyperlink, number of turns, and participants (presenter and readers). If different readers commented on one presentation, readers were labelled as Reader 1, Reader 2, Reader 3, and so on, to track the different discussions that took place.

3.3. Analytical procedure

The TCDS documents were uploaded to the qualitative data software ATLAS.ti. version 8.4.5.⁴ The codification of the corpus started with an inductive approach based on a close reading of the corpus tweets. To assure the reliability of the annotation process, the author used the memo and code description options of ATLAS.ti, which allow describing in detail coding procedures and decisions that help to guarantee consistency during the labelling process (Krippendorff 2004; Paulus 2022). The coding system was revised and redefined in three cycles to reach a saturation point of codes and carried out in three-time intervals to guarantee the validity of the codification (Saldaña 2009).

The coding cycle began with the identification and description of the technical and discursive features shaped by the medium (e.g., Twitter affordances and constraints, digital discourse features). In particular, I focused on multimodal semiotic resources (Orpin 2019; Luzón and Albero-Posac 2020; Luzón 2023), Twitter formal elements (Luzón 2023; Tardy 2023), and linguistic features common in digital discourse (Mauranen 2013; Tagg 2015; Zappavigna 2017; Luzón and Albero-Posac 2020). Regarding the latter, previous studies on face-to-face discussion sessions have analysed non-linguistic resources such as gestures, facial expressions, loudness, or laughter (Wulff et al. 2009; Querol-Julián and Fortanet-Gómez 2012). Transferring these features into the digital medium can be done through visual resources (emojis, smileys), punctuation (exclamation marks, capitalisation), or characteristic features of spoken dialogue (addressing interlocutors by their first names, interjections, laughter, or lengthened vowels). Lastly, I examined features of interpersonality as established in Hyland’s (2005) stance and engagement framework, to pinpoint what strategies were used by presenters and readers to interact with one another (Querol-Julián and Fortanet-Gómez 2012; Orpin 2019; Luzón and Albero-Posac 2020; Luzón 2023).

The communicative functions of tweets were also coded. For readers’ tweets, i.e., questions and comments, I followed Xu’s (2022a) taxonomy of questions (fact-seeking, opinion-seeking, justification-seeking, suggestion-making, and comment) for an initial overview of communicative functions. Regarding questions, 95 percent were contextualised questions, which meant that in addition to the question itself, other moves such as announcing the question, greeting, praising, or sharing some personal information relevant to the situation before posing the question were needed, a similar situation to what others had previously noted in their analysis of face-to-face discussion sessions (Querol-Julián and Fortanet-Gómez 2014; Xu 2022a). After a reiterative process of re-reading tweets and a redefinition of codes, 32 communicative functions were identified and classified into three main communicative purposes: knowledge construction, community building, and self-promotion. Table 2 summarises the categories and codes of the coding system.

Category	Codes
Multimodal resources	Emojis, gifs, images, smileys, videos.
Twitter formal elements	Embedded tweets, hashtags, hyperlinks, mentions.
Digital discourse features	Abbreviations, capitalisation, contractions, intensifiers (adverbs, repetition of words/symbols), exclamation marks, interjections, laughter, lengthened vowels.
Interpersonality	Stance: Self-mentions (first person pronouns, possessives), hedges (modal verbs and conversational hedges e.g., just, a little bit), attitude markers (evaluative adjectives, verbs).
	Engagement: Reader mentions (second person pronouns, possessives, vocatives), personal asides.
Knowledge construction	Acknowledging collaboration, acknowledging limitations, agreeing with a previous idea, asking for feedback, discussing an idea, exemplifying, explaining content, making requests, making suggestions, offering a neutral statement, referring to previous studies, requesting clarification, requesting an opinion, seeking factual information, seeking justification, sharing resources.
Community building	Addressing the reader, apologising, appraising the presenter’s work, conveying gratitude, down-toning, engaging in humour, expressing politeness, expressing strong feelings, greeting, keeping in touch after the conference, sharing personal information, sharing research interests.
Self-promotion	Expressing significance, promoting one’s outputs, raising awareness, referring to future work.

Table 2: Description of the coding system

4. Results

The results show that the most frequent communicative purposes of TCDS tweets are community building and knowledge construction. These findings are reported in Section 4.1., where communicative functions and rhetorical strategies are analysed in relation to the different turn-types of TCDSs. Section 4.2. reports on the digital medium-related characteristics of TCDSs.

4.1. Communicative functions and discursive realisations of turns

4.1.1. Comments

Comments often take the form of statements written by the reader and are related to an interpersonal or community-building dimension, hence, granting more importance to building rapport and interpersonal relations than to knowledge construction (Table 3).

Community building (N=208)
Communicative Function	Frequency
Appraising the presenter’s work	83
Addressing the reader	22
Conveying gratitude	21
Engaging in humour	20
Expressing strong feelings	20
Sharing personal information	15
Sharing research interests	13
Keeping in touch after the conference	7
Apologising (for a mistake)	3
Greeting	2
Down-toning	1
Expressing politeness	1
Knowledge construction (N=99)
Communicative Function	Frequency
Offering a neutral statement	40
Making suggestions	15
Referring previous studies	9
Explaining content	8
Exemplifying	7
Sharing resources/outputs	6
Agreeing with a previous idea	4
Seeking factual information	4
Acknowledging collaboration	2
Making requests	2
Acknowledging limitations	1
Asking for feedback	1
Self-promotion (N=26)
Communicative Function	Frequency
Expressing significance	11
Promoting one’s outputs	9
Raising awareness	6

Table 3: Frequencies of communicative functions in comments.

Comments conveyed the following functions: a) appraising the presenter’s work, b) offering a neutral statement, c) addressing the reader, d) conveying gratitude, e) engaging in humour, and f) expressing strong feelings. All of them work to establish rapport and a positive evaluation of the conversation that takes place. Examples (1–6)⁵ illustrate the different functions:

1. Super interesting presentation! Thank you! I’ve never thought about memes as giving advice before, but it makes sense. I’ll be keeping my eyes out for that now. (TCDS6_Reader3)
2. I’ve heard my dad & uncle (both from Michigan, USA but raised by two Appalachian English–speaking parents) say things like “They wanted to get married real quick” to mean “they wanted to get married in a short amount of time.” For me though I can’t do this with postverbal “quick” (TCDS14_Reader1)
3. Really enjoyed this Martin! So clear and fun :) always a pleasure to read your stuff, greetings from NZ (TCDS26_Reader5)
4. Thanks for the paper! and the refs! (TCDS23_Reader1)
5. Meow Viry Much Madame ! We potitchats from France are realy proud /20 (TCDS45_Reader1)
6. Awwww we lost! we (renov) were winning at one point. (TCDS26_Reader1)

Comments correlate with appraising the presenter’s work at the beginning of the tweet (1–4) and sharing personal information and research interests (1–2). At the textual level, a high frequency of self-mentions through the first-person pronouns I (1–2) or we representing a group is found (5–6). Subject omission is also observed (3), as part of a more conversational register, which is also noticed in other linguistic digital discursive features such as the use of exclamation marks, especially after thanking or requesting, abbreviations and contractions, spelling mistakes, letter repetition, interjections, or the use of vocatives.

Evaluative language, in particular positive evaluation, is a common trait in TCDSs through the use of intensifiers (super, really) and adjectives describing the presentation’s contents, as shown in (1), (3), and (5). The use of these strategies creates a close bond between the presenter and the reader, both relying on politeness strategies to create rapport by positively appraising the presenter, addressing the reader directly, and conveying gratitude (3–4). These communicative functions reflect a focus on the person rather than on the (scientific) content. Similarly, the use of humour in (5) is another strategy that can include inside-group jokes and references to shared interests between the presenter and the reader. As found by Wulff et al. (2009) in their analysis of laughter in conference discussion sessions, laughter and humour tend to be present to soften potential criticism or requests, or to break the ice at the beginning of a conversation.

4.1.2. Questions

Table 4 shows the distribution of communicative purposes with their corresponding communicative functions in questions. The main purpose of questions is to construct and exchange knowledge, closely followed by community building.

Community building (N=121)
Communicative Function	Frequency
Appraising the presenter’s work	40
Conveying gratitude	15
Expressing politeness	15
Addressing the reader	11
Expressing strong feelings	6
Greeting	5
Sharing research interests	5
Engaging in humour	3
Down-toning	1
Knowledge construction (N=127)
Communicative Function	Frequency
Seeking factual information	48
Requesting an opinion	25
Making requests	13
Making suggestions	12
Requesting clarification	8
Referring previous studies	6
Exemplifying	5
Sharing resources/outputs	5
Seeking justification	4
Agreeing with a previous idea	1
Self-promotion (N=1)
Communicative Function	Frequency
Referring future work	1

Table 4: Frequencies of communicative functions in questions

As far as questions are concerned, the main communicative functions they fulfil are four: a) seeking factual information, b) appraising the presenter’s work, c) seeking opinion, and d) expressing politeness. Two out of the three most frequent functions coincide with Xu’s (2022a) taxonomy of conference questions as illustrated in:

7. Are there books that use the same font for both? (TCDS3_Reader1)
8. Beautiful graph! What did you use to do that? (TCDS5_Reader1)
9. Thank you for this talk! We agree, I think. We looked into German and Dutch a little and wondered about the distinction between the subordinating and coordinating becauses. Do you have any thoughts on that? (TCDS55_Reader1)

Example (7) illustrates a straight question that is purely fact-seeking, but examples (8) and (9) show how readers prefer to contextualise the question before making a request. For instance, questions are introduced first by praising the presentation or a specific part of the presentation with positive evaluative adjectives (8) or by congratulating and sharing some research interests that position the reader at the same level as the presenter in terms of knowledge (9). Likewise, when requesting an opinion from the presenter, in addition to addressing the presenter directly with the pronoun you, readers often appraise their work and convey gratitude using the same linguistic resources and standard formulaic politeness strategies (full grammatical sentences, polite requests, hedging) before posing a question that could be interpreted as threatening to the presenter’s expertise.

4.1.3. Responses

The main communicative purposes of response tweets by the presenter are knowledge construction and community building. Regarding the third communicative purpose, self-promotion, it has the highest occurrence in the response category (Table 5).

Community building (N=158)
Communicative Function	Frequency
Conveying gratitude	70
Addressing the reader	19
Appraising the presenter’s work	14
Keeping in touch after the conference	14
Engaging in humour	9
Expressing strong feelings	8
Sharing personal information	6
Greeting	5
Apologising (for a mistake)	4
Sharing research interests	4
Down-toning	3
Expressing politeness	2
Knowledge construction (N=197)
Communicative Function	Frequency
Explaining content	83
Acknowledging limitations	25
Agreeing with a previous idea	25
Exemplifying	20
Referring previous studies	14
Sharing resources/outputs	14
Requesting clarification	6
Making requests	5
Acknowledging collaboration	2
Asking for feedback	2
Requesting an opinion	1
Self-promotion (N=30)
Communicative Function	Frequency
Promoting one’s outputs	16
Referring future work	13
Expressing significance	1

Table 5: Frequencies of communicative functions in responses

Presenter response tweets realise five communicative functions in the data: a) explaining content, b) conveying gratitude, c) agreeing with previous ideas, d) acknowledging limitations, and e) exemplifying, as shown in:

10. Yes, we looked at all loanwords that occurred at least five times across our corpus, regardless of their meaning, and we do find loanwords within the same text that are not semantically related, e.g. paua (shell) and aroha (love) :) (TCDS5_Presenter)
11. Thank you! Yes, I think they all derive from the original POSS.2SG along the path: possessive > salient > anaphoric > proprial Although, I'm not exactly sure, how the last link works. I hope to get a chance to present these hypothesis at #SLE2021 (TCDS21_ Presenter)
12. I am definitely not! I fully agree with you and thank you for pointing this out. This was just to make it easy for people to know what I was talking about in the first tweet. We were limited to six! 😳 (TCDS45_ Presenter)

Aligning with the nature of responses in face-to-face discussion sessions, their main communicative function in TCDSs is explaining by elaborating on content to answer questions (10) and (11). Explanations often appeared in combination with exemplification, a strategy used by presenters to illustrate abstract concepts. Similar to questions and comments, conveying gratitude by thanking the other person (for either reading the presentation or posing a question) could be considered an obligatory function in view of its frequent use in responses. As part of a friendly and polite environment, many responses agreed with previous comments by readers (10–12). These functions are realised with exclamation marks after thanks or thank you to stress friendliness and enthusiasm, emojis and smileys, evaluative language (I fully agree), and the use of first person-pronouns that make the presenter’s voice visible (10–12).

Another significant function is the authors’ acknowledgment of the limitations of their research by hedging (11) or down-toning, as in (12), where the presenter acknowledges that the presentation content has been simplified because of space constraints. This positioning shows presenters not as knowledge holders but rather as participants in the knowledge construction process.

Lastly, the self-promotion communicative purpose, even though it occurs less frequently than the community building and knowledge construction purposes, occurs most frequently in responses, especially through the function of promoting one’s publications and outputs. This might result from the fact that presenters are expected to provide references for their presentations’ contents.

4.1.4. Follow-up conversations

When the conversation between reader and presenter continued after the presenter’s response, community building and knowledge construction continued to be the most relevant purposes of longer interactions. As shown in Table 6, the main communicative functions of tweets were: a) conveying gratitude, b) agreeing with a previous idea, c) explaining content, d) appraising the presenter’s work, e) expressing strong feelings, and f) keeping in touch after the conference.

Community building (N=161)
Communicative Function	Frequency
Conveying gratitude	43
Appraising the presenter’s work	29
Expressing strong feelings	18
Keeping in touch after the conference	17
Engaging in humour	16
Sharing personal information	13
Down-toning	7
Expressing politeness	6
Sharing research interests	5
Apologising (for a mistake)	4
Addressing the reader	3
Knowledge construction (N=155)
Communicative Function	Frequency
Agreeing with a previous idea	35
Explaining content	31
Referring previous studies	13
Acknowledging limitations	12
Discussing an idea	12
Sharing resources/outputs	11
Exemplifying	10
Making requests	9
Seeking factual information	8
Making suggestions	6
Offering a neutral statement	3
Requesting an opinion	2
Requesting clarification	2
Asking for feedback	1
Self-promotion (N=13)
Communicative Function	Frequency
Promoting one’s outputs	7
Referring future work	5
Expressing significance	1

Table 6: Frequencies of communicative functions in follow-up conversation

As follow-up conversations consist of several tweets between readers and presenters, in particular further comments and responses, communicative functions such as conveying gratitude and positively appraising the presenter’s work are commonly found. With this turn-type, the conversation topic is expanded, so agreeing with previous ideas presented in the tweets is a frequent function. Agreeing is also used as a positive politeness strategy to foster bonds between readers and presenters. However, whenever the reader insists on the topic, the use of hedges (I was just curious) and other polite strategies (I was wondering if you could have, but maybe this is already moving too far) as well as relying on specific resources (I’ll quickly look at my data, I’m going to look up articles) are frequent to justify their questions and answers, as illustrated in (13) and (14):

13. Reader2: Great thread!! I was just curious about what you make of l.14? 😄
    Presenter: Thank you! :) I consider l. 14 has a confirmation and acceptance of the proposed other-increments. I find it interesting how Anna uses overt dependent syntax (“che”) so her talk is dependent on Paolo’s... but Paolo’s turn l. 14 is an independent clause. ;)
    Reader2: Thank you! Yeah, I was wondering if you could have cases in which the acceptance is also designed as dependent sort of recycling the increment. But maybe this is already moving too far from your point here hehe. Anyway, great job!
    Presenter: No, but it’s a great question. It allows me to think about the notion of recycling/repeating in relation to acceptance/confirmation of a candidate! I’ll quickly look at my data and come back to you! :D thank you so so much! (TCDS39_Follow-upConversation)

These polite strategies are interwoven with digital discourse features like exclamation marks, laughter, repetitions, and contractions to seem friendlier. In a similar vein, emojis and smileys are more frequent in this turn once that contact has already been established between users.

14. Reader2: This is super interesting, thank you! In my research I look at the link between acquisition and language change. Would you say this could link up with a cue-based approach to change [...] ½
    Reader2: i.e children are sensitive to prosodic cues in their input and this can cause frequency changes for linguistic forms, leading to overall language change?
    Presenter: I don’t know much about the link between language acquisition and change, though I find it a very interesting subject. But I would guess yes, since prosodic cues are so important for language acquisition, they could also influence language change through changes in input cues.
    Reader2: I wonder if there’s any evidence out there that’s shown change occurring in synchronic acquisition data. I know Marit Westergaard works on syntactic cues a lot, so was just wondering if it extended to prosody! Thanks!
    Presenter: Thank you for the question! I wish I could help you more, but unfortunately I can't think of any study on this subject. If I do think of something, I’ll let you know :)
    Reader2: No problem at all! Thanks for sharing your research. Likewise, I think I’m now going to look up articles related to prosodic cues and change so will let you know 😃 (TCDS34_follow-upConverstation)

Likewise, expressing strong feelings through evaluative language (e.g., great, cool, I’d love to), and using adverbial intensifiers (super, so), reader mentions (you, your interest) or conditional sentences (if I do think of…, I wish I could…) might be associated with various communicative functions such as keeping in touch after the conference, exchanging resources, or discussing ideas. All in all, participants try to come across as friendly and supportive.

4.2. Digital discourse features of TCDSs

In view of TCDSs taking place on a digital platform, communicative practices from physical discussion sessions can be digitally remediated or new practices might emerge resulting from Twitter’s affordances and constraints. Table 7 shows the frequency and distribution of Twitter formal elements (embedded tweet, hashtag, mention), hyperlinking, and multimodal assemblage of semiotic resources (image, gif, emoji/smileys). Total frequencies are broken down by turn-type. The features that stand out the most in TCDSs are emojis/smileys, mentions, and hyperlinks.

Features	Comment	Question	Response	Follow-up	Total
Embedded tweet	1	1	2	1	5
Hashtag	12	3	2	1	18
Mention	24	8	5	1	38
Hyperlink	2	4	9	5	20
Total Twitter formal elements	39	16	18	8	81
Emoji/smiley	47	11	58	47	163
Gif	0	0	0	1	1
Image	1	0	0	1	2
Total multimodal elements	48	11	58	49	166

Table 7: Distribution of Twitter formal elements and multimodal elements by turn-type

In opposition to TCP, which are heavily loaded with images (Villares 2023a), visual elements such as images and gifs are scarce in TCDSs. Only emojis and smileys are used frequently by both readers and presenters. As identified in previous sections, the main function of emojis is to show the attitude of the reader or presenter, which corresponds with either a positive evaluation that helps to create a sense of closeness and friendliness, as in (10) and (12–14), or to express concerns when acknowledging limitations or explaining content, (15–16):

15. Possibly, but these stigmatized variants are mostly associated with rural areas. Perhaps they have been lost in those rural areas, or perhaps they are not as stigmatized anymore because of migration. That’s something we wonder now 🤔🤷‍♀️ ... It’s too small a study to know for sure at this point! (TCDS4_Presenter_Follow-upConversation)
16. I don’t know if reviewer #2 will accept our conclusions on the basis of a twitter poll, but I’m sure glad it’s in line with our theory. 😅 Across languages, the suprasegmental rules of clipping vary substantially. (TCDS26_Presenter_Follow-upConversation)

Regarding the praising and conveying gratitude functions, sometimes, tweets with no text, only emojis (e.g., 👏👏👏, 👌🏻, 😻), could be found. They frequently represented a paralinguistic feature common in the discussion session, i.e., clapping, agreeing, or smiling. In other cases, emojis appear alongside text, reinforcing or enhancing the tweet’s communicative function.

Moving on to Twitter’s specific formal elements, mentions were the preferred resource. Like vocatives, mentions tend to appear at the beginning of the tweet, often creating a sense of proximity, and allowing immediate interaction because it notifies and explicitly addresses other users. Mentions are followed by communicative functions such as expressing strong feelings, expressing gratitude, or praising the presenter’s work as illustrated in (17):

17. Trabalho maravilhoso, @ NicolaDaly18! A multimodalidade tem se mostrado uma excelente aliada no ensino-aprendizagem de línguas. (Wonderful study, @NicolaDaly18! Multimodality has proved to be an excellent ally in the learning and teaching of languages) (TCDS3_Reader2_comment)

Other uses of mentions can refer to a presenter naming co-authors (18), calling out the author of a resource that could be useful within the discussion (19), or informing a third person of the existence of the presentation (20):

18. Co-author on this work is @elles_belles (who I didn’t tag because she never tweets haha) (TCDS31_Presenter_comment)
19. @uhlon dohlenko wrote a paper about anglophone lolspeak a loooong time ago! (I want to say “before it was cool”, but I guess that was actually at the height of its popularity?) (TCDS45_Reader3_comment)
20. @pbcardoso, see this! (TCDS36_Reader3_comment)

Hyperlinks are another digital affordance that fulfils the communicative function of sharing resources and promoting one’s outputs or publications (e.g., links to a paper’s DOI and repositories) and sharing resources (e.g., software, code, websites) by both presenters and readers. The remaining Twitter technical elements are hashtags, which are mainly used to relate the content of tweets to the conference (e.g., #SLE2021) or for humoristic purposes (e.g., #SauronEye, #SavetheGricean). Embedded tweets work as referencing tools so that participants can point to specific information mentioned during presentations to share outputs/resources or as promotional tools that increase the presentation’s visibility when it is shared in other discussions.

5. Conclusion

This study has explored how TCDSs remediate the face-to-face academic conference discussion session digitally. Both genres have a similar sequential organisation beginning with a comment or question posed by a reader and finishing with the presenter’s response (Querol-Julián and Fortanet-Gómez 2014). However, the digital genre introduces some novelty, particularly when interaction exceeds two turns (i.e., tweets). This turn-type, which I labelled ‘follow-up conversation’, consists of both readers and presenters elaborating on their answers and expanding the conversation to areas of knowledge construction, community building, and self-promotion. The automatic comment, which refers to promotional software-generated tweets, was also considered a medium-related turn. Regarding readers’ turns, while TCDS questions can be either straightforward or contextualised (Xu 2022a), most tweets were contextualised, which fostered a bond between participants by drawing on different community-building communicative functions.

Regarding the communicative purposes of TCDSs, participants engage in conversation to establish interpersonal relationships (community building), exchange knowledge (knowledge construction), and to a lesser extent, self-promotion. Hence, TCDS participants engage in conversations with similar communicative purposes as users of other digital genres such as science blogs, academic conference tweets, or tweetorials (Mauranen 2013; Luzón and Albero-Posac 2020; Tardy 2023). Community-building communicative functions are means to establish interpersonal relationships between presenters and readers in a positive and polite manner. Compared to face-to-face conferences, paralinguistic strategies (e.g., gestures, body language, facial expressions) are remediated in the digital medium with the adoption of informal digital discourse. By using vocatives, exclamation marks, emojis, or evaluative language, participants create rapport and a friendly environment. Knowledge-construction communicative functions commonly include an exchange of specific information and opinions that should be explained and justified with examples, data, or references. This often occurs in follow-up conversations because they grant more space to delve into the topics, while face-to-face discussions tend to give hush answers due to time constraints. Moreover, Twitter allows participants to use its affordances, i.e., hyperlinking and multimodality, to give richer answers and move discussions forward.

In opposition to what previous academic Twitter research suggests (Büchi 2016; Lee et al. 2017; Côté and Darling 2018; or Mehlenbacher 2019), self-promotion is not a relevant communicative purpose for participants during the discussion. This finding contrasts with the TCP, the previous genre to the TCDS. In TCPs, presenters focus on knowledge construction and self-promotion reflected on the use of discursive strategies such as questions, informative images, or semantic hashtags to signal key terms to make the presentations attractive and informative (Villares 2023a, 2023b). In the case of TCDSs, however, there is a focus on community-building and networking, therefore, relying on a conversation style characterised by emojis/smileys, exclamative sentences, and evaluative language. These discursive features are enhanced by the digital medium and imitate both linguistic and paralinguistic features of face-to-face interaction.

This study presents some limitations that should be commented on. The study’s data come from a small corpus that should be expanded to include a larger corpus of academic tweets, either from TCDS or other Twitter-related genres such as publication-promoting tweetorials or other academic-related threads (Luzón 2023; Tardy 2023). Likewise, the analytical framework could be applied to other digital genres that promote interaction among diversified audiences such as Reddit forums or citizen science websites to test its efficacy in analysing science communication practices. A second limitation refers to the fact that the data analysis was carried out by just one person. Even though contingency measures were implemented to ensure consistency and the validity of results with ATLAS.ti tools, it is advisable for future studies to involve more researchers who could ensure high inter-rater reliability agreement levels. Thirdly, from a methodological perspective, the discursive analysis of tweets could have been complemented with interviews or questionnaires to some of the presenters and readers to validate the study’s results. This action would have shed light on the participants’ actions and intentions, and it might open a future avenue for research.

Finally, as this paper has described digital communicative practices of international academic communities, the findings may have some pedagogical implications. Nowadays researchers find themselves in a paradigm where science needs to be communicated in a transparent, accessible, and engaging way, yet few opportunities to learn and develop this skill are offered by their institutions. To achieve this goal, it is crucial to understand how digital science communication happens so that research-based training shows scholars how to share scientific knowledge, prompt discussions, or foster collaboration among diversified audiences in the new digital genres. Hence, this study contributes to the current research on the identification of new communicative practices within the framework of digital genre analysis and the importance of social media for community building and knowledge construction.

References

Belcher, Diane D. 2023. Digital genres: What they are, what they do, and why we need to better understand them. English for Specific Purposes 70: 33–43.

Büchi, Moritz. 2016. Microblogging as an extension of science reporting. Public Understanding of Science 26/8: 953–968.

Côté, Isabelle M. and Emily S. Darling. 2018. Scientists on Twitter: Preaching to the choir or singing from the rooftops? FACETS 3/1: 682–694.

Darling, Emily S., David Shiffman, Isabelle M. Côté and Joshua A. Drew. 2013. The role of Twitter in the life cycle of a scientific publication. Ideas in Ecology and Evolution 6: 32–43.

Hyland, Ken. 2005. Stance and engagement: A model of interaction in academic discourse. Discourse Studies 7/2: 173–192.

Hyland, Ken. 2009. Academic Discourse: English in a Global Context. New York: Continuum.

Konzett, Carmen. 2012. Any Questions? Identity Construction in Academic Conference Discussions. Berlin, Boston: De Gruyter Mouton.

Krippendorff, Klaus. 2004. Content Analysis: An Introduction to its Methodology. Thousand Oaks: Sage Publications.

Lee, Mi Kyung, Ho Young Yoon, Marc Smith, Hye Jin Park and Han Woo Park. 2017. Mapping a Twitter scholarly communication network: A case of the association of internet researchers’ conference. Scientometrics 112/2: 767–797.

Luzón, María José. 2023. Multimodal practices of research groups in Twitter: An analysis of stance and engagement. English for Specific Purposes 70: 17–32.

Luzón, María José and Sofía Albero-Posac. 2020. ‘Had a lovely week at #conference2018’: An analysis of interaction through conference tweets. RELC Journal 51/1: 33–51.

Luzón, María José and Carmen Pérez-Llantada. 2022. Digital Genres in Academic Knowledge Production and Communication: Perspectives and Practices. Bristol: Multilingual Matters.

Mauranen, Anna. 2013. Hybridism, edutainment, and doubt: Science blogging finding its feet. Nordic Journal of English Studies 12/1: 7–36.

Mehlenbacher, Ashley Rose. 2019. Science Communication Online: Engaging Experts and Publics on the Internet. Columbus: The Ohio State University Press.

Miller, Carolyn R. 1984. Genre as social action. Quarterly Journal of Speech 70: 151–167.

Orpin, Deborah. 2019. #Vaccineswork: Recontextualizing the content of epidemiological reports on Twitter. In María José Luzón and Carmen Pérez- Llantada eds. Science Communication on the Internet: Old Genres Meet New Genres. Amsterdam: John Benjamins, 173–194.

Paulus, Trena. 2022. Digital tools for digital discourse analysis. In Camilla Vásquez ed. Research Methods for Digital Discourse Analysis. New York: Bloomsbury Publishing, 115–137.

Pérez-Llantada, Carmen, Olga Abián, Cristina Cadenas-Sánchez, Oana Carciu, Jesús Clemente-Gallardo, Idoia Labayen, Bienvenido León, Maria Carmen Erviti, Alfonso Ollero, Maddi Oses Recalde, Diego Rivera, Alberto Vela, Adrian Velazquez-Campoy, Rosana Villares and Ana Cristina Vivas Peraza. 2022. Digital Science: Sustainable, Transformative and Transversal. Final Report. Mendeley Data V1. https://doi.org/10.17632/2yv5brwxg5.1

Querol-Julián, Mercedes and Inmaculada Fortanet-Gómez. 2012. Multimodal evaluation in academic discussion sessions: How do presenters act and react? English for Specific Purposes Journal 3/4: 271–283.

Querol-Julián, Mercedes and Inmaculada Fortanet-Gómez. 2014. Evaluation in discussion sessions of conference presentations: Theoretical foundations for a multimodal analysis. Kalbotyra 66: 77–98.

Räisänen, Christine. 2002. The conference forum: A system of interrelated genres and discursive practices. In Eija Ventola, Celia Shalom and Susan Thompson eds. The Language of Conferencing. Berlin: Peter Lang, 69–93.

Rowley-Jolivet, Elizabeth and Shirley Carter-Thomas. 2005. The rhetoric of conference presentation introductions: context, argument and interaction. International Journal of Applied Linguistics 15/1: 45–70.

Saldaña, Johnny. 2009. The Coding Manual for Qualitative Researchers. Thousand Oaks: Sage Publications.

Swales, John M. 1990. Genre Analysis: English in Academic and Research Settings. Cambridge: Cambridge University Press.

Tagg, Carolyne. 2015. Exploring Digital Communication: Language in Action. New York: Routledge.

Tardy, Christine M. 2023. How epidemiologists exploit the emerging genres of Twitter for public engagement. English for Specific Purposes 70: 4–16.

Villares, Rosana. 2023a. Twitter conference presentations: A rhetorical and semiotic analysis of an emerging digital genre. ELIA: Estudios de Lingüística Inglesa Aplicada 22: 125–167.

Villares, Rosana. 2023b. Exploring rhetorical strategies of stance and engagement in Twitter conference presentations. ESP Today 11/2: 280–301.

Walter, Stephanie, Ines Lörcher and Michael Brüggemann. 2019. Scientific networks on Twitter: Analyzing scientists’ interactions in the climate change debate. Public Understanding of Science 28/6: 696–712.

Webber, Patrick. 2002. The paper is now open for discussion. In Eija Ventola, Celia Shalom and Susan Thompson eds. The Language of Conferencing. Berlin: Peter Lang, 227–253.

Wulff, Stefanie, John M. Swales and Kristen Keller. 2009. “We have seven minutes for questions”: The discussion sessions from a specialized conference. English for Specific Purposes 28: 79–92.

Xu, Xiaoyu. 2022a. A genre-based analysis of questions and comments in Q&A sessions after conference paper presentations in computer science. English for Specific Purposes 66: 63–76.

Xu, Xiaoyu. 2022b. Differences between novice and experienced academics in their engagement with audience members in conference Q&A sessions. Journal of English for Academic Purposes 60: 101188. https://doi.org/10.1016/j.jeap.2022.101188

Xu, Xiaoyu, Jerois Gevers and Luca Rossi. 2023. “Can I write this is ableist AF in a peer review?”: A corpus-driven analysis of Twitter engagement strategies across disciplinary groups. Ibérica 46: 207–236.

Zappavigna, Michelle. 2017. Twitter. In Christian Hoffmann and Wolfram Bublitz eds. Pragmatics of Social Media. Berlin: De Gruyter Mouton, 201–224.

Appendix 1

1. Citing linguistic data: The Tromsø Recommendations:
   https://twitter.com/superlinguo/status/1335011220152229888
2. What’s in a name?
   https://twitter.com/sheeli3/status/1335015044745191431
3. The Linguistic Landscape of Bilingual Picturebooks:
   https://twitter.com/NicolaDaly18/status/1335023291937947648
4. Value judgments associated with allophones of alveolar tap and trill:
   https://twitter.com/porraschaver/status/1335029874285633540
5. Exploring Loanword Networks:
   https://twitter.com/TryeDavid/status/1335037615326674944
6. Tempering and Aligning Advice-Giving Through Memes:
   https://twitter.com/Ling_Lass/status/1335045068701503494
7. Place identity & co-occurrence in Northern Maine:
   https://twitter.com/Katharina_Pabst/status/1335049287097589761
8. The demise of impersonal constructions:
   https://twitter.com/chao_noelia/status/1335052686442553346
9. Determinatives are far from pronouns in English:
   https://twitter.com/brettrey3/status/1335060090941018112
10. Emoji based reactions to the Said Construction:
    https://twitter.com/AliciaStevers/status/1335064270833405952
11. #AboriginalEnglish: BE LIKE, stability and change:
    https://twitter.com/CelesteRLouro/status/1335067954019323909
12. Perception of American English pure vowels:
    https://twitter.com/NaimAfshar/status/1335071401934467074
13. Interactionally situating the power scream:
    https://twitter.com/EMdoesCA/status/1335075479506857986
14. Why are we *quick* to point out, but not *fast*?
    https://twitter.com/demeco_project/status/1335079000155295747
15. What the Italian subjunctive actually means:
    https://twitter.com/salviodigesto/status/1335082822600634368
16. Empirical methods for describing TAM:
    https://twitter.com/AnaKrajinovic1/status/1335086658962780161
17. Positional Preference of Emotion Phrase in Hindi:
    https://twitter.com/spandan_ju/status/1335090484667039746
18. We don’t agree (only) upwards:
    https://twitter.com/Andraas/status/1335094108436750337
19. The F2 Robot Interaction System:
    https://twitter.com/f2robot/status/1335097926629126148
20. Acquisition of syntactic negation & NC:
    https://twitter.com/samrinice/status/1335101660633440256
21. Kazym Khanty -en: 2SG possessive—>proprial article:
    https://twitter.com/SK_Mikhailov/status/1335105393861808128
22. Cracking stereotypes the ling of discourse markers:
    https://twitter.com/AichaBelkadi/status/1335109284170969088
23. Lexical classification of Tupí-Guaraní languages:
    https://twitter.com/fthorstensen/status/1335113053571080193
24. The way Spanish and Basque think about causality:
    https://twitter.com/AndreaArioBizar/status/1335116703517331456
25. Social influence on negation in Early Modern Dutch:
    https://twitter.com/leviremijnse/status/1335120535303430144
26. How to clip words in English:
    https://twitter.com/hilpert_martin/status/1335124283526422529
27. The QUD in quantity judgments:
    https://twitter.com/kerbach2/status/1335129528524529664
28. Intensifiers across social media:
    https://twitter.com/tschfflr/status/1335132021216174082
29. Gaze-selection & syntax in multiperson interaction:https://twitter.com/cal_virgi/status/1335135674039754754
30. Variation in framing of real-world events:
    https://twitter.com/gossminn/status/1335139354063347713
31. Because-X and because-ellipsis: A comparison:
    https://twitter.com/linguistlaura/status/1335143470684663810
32. who gives what to whom, and how do we know:
    https://twitter.com/evaeva_z/status/1335147032164626433
33. Processing linguistic variation:
    https://twitter.com/rkofreitag/status/1335171551025565700
34. Children use prosody for sentence disambiguation:
    https://twitter.com/KolbergLeticia/status/1335177147401592832
35. Feedback as the main mechanism of L2 processing:
    https://twitter.com/AmandaP27090148/status/1335196268042260481
36. Multimodal perception of Brazilian Portuguese:
    https://twitter.com/Lumamirand/status/1335203546669658112
37. Matrix Language in the Code-Switching in Children:
    https://twitter.com/KFascinettoZ/status/1335208175130198020
38. Efficient coding: Passive and dative alternations:
    https://twitter.com/haspelmath/status/1335212273250562048
39. How priming in bilinguals leads to language change:
    https://twitter.com/EvangeliaAdamou/status/1335218667580252160
40. Bilingual mixed NPs: speech data vs. models:
    https://twitter.com/MixedNPs/status/1335222452826165249
41. Pragmatic and discursive mechanisms in headlines:
    https://twitter.com/daniel__pascual/status/1335229942758510592
42. What counts as alternating passive constructions?
    https://twitter.com/marciamv2/status/1335233794585092097
43. Gricean Secrets:
    https://twitter.com/anthony69848604/status/1335237503775870983
44. Cats of Twitter:
    https://twitter.com/BerLinguistin/status/1335241599715074051
45. National language literacy lessons in The Gambia:
    https://twitter.com/clydeancarno/status/1335245117964300288
46. Ideologies in a university linguistic landscape:
    https://twitter.com/ruiality/status/1335248929022238720
47. The Decline of V2 in the History of English:
    https://twitter.com/sophiewhittle95/status/1335260153877237760
48. Translanguaging lens in deaf education:
    https://twitter.com/AryaneSNogueira/status/1335279092124569601
49. The Changing Language of the Climate Change Debate:
    https://twitter.com/RDFT58485932/status/1335301693513273347
50. Be that as it may: The Unremarkable Trajectory of the North American English Subjunctive:
    https://twitter.com/mizlinguist/status/1335309295215390721
51. Teasing on Twitter: An analysis of Donald Trump’s tweets:
    https://twitter.com/lillapszabo/status/1335312990174896136
52. Frame Semantics & Multimodal Machine Translation:
    https://twitter.com/viridiano/status/1335286721269919745
53. One Too Many Plural(s): Taglish Code-Switching:
    https://twitter.com/petertorres/status/1335320655143731201
54. A Classless Analysis of Italian Nouns:
    https://twitter.com/ulfsbjorninn/status/1335354516275912704
55. Because X: Now I’m an ellipsis and now I’m not:
    https://twitter.com/TeapotLinguist/status/1335365957401907200

Notes

¹ This study was supported by the project Digital Genres and Open Science (PID2019-105655RB-I00 MCIN/AEI 10.13039/501100011033) funded by the Spanish Ministry of Science and Innovation and the Spanish Agency for Research and the Government of Aragon (H16_23). It is also a contribution to the Erasmus + Project Digital Language and Communication Training for EU Scientists (DILAN), co-funded by the European Commission (2022-1-ES01-KA220-HED-000086749). This publication reflects the views only of the author, and the Commission cannot be held responsible for any use which may been made of the information contained therein. [Back]

² A list with the presentation titles and links can be found in Appendix 1. [Back]

³ https://abralin.org/es/evento/linguistweets-3/ [Back]

⁴ https://atlasti.com/ [Back]

⁵ Examples are verbatim transcriptions of tweets. Tweets not written in English include the translation in brackets. [Back]