Interests and feelings of Spanish teachers on twitter during Covid-19

Abstract

The diffusion of Covid-19 imposed the confinement of the entire Spanish population between March and June 2020. An exploratory study of the evolution of teachers’ interests and feelings in social networks is carried out. For this purpose, three key periods have been differentiated: pre-pandemic (January 2018-February 2020), confinement and state of alarm (March 2020-May 2021), and post-pandemic (June 2021-March 2022). We analyzed 9462 tweets from faculty communities on Twitter using public data mining, sentiment analysis, and semantic content analysis. The results highlight a significant increase in teacher interventions on Twitter and an increase in emotional content during the confinement in both positive and negative sentiments, closely related to the use of emojis. During the pandemic period, new topics of interest appeared related to the evolution of COVID-19 in Spain and the safety of the return to the classroom. Topics dealing with digital tools and innovative methodologies appeared with different relevance in all the periods analyzed.

Keywords

Twitter Spanish teachers public opinion content analysis sentiment analysis covid-19

Introduction

The rapid and unexpected spread of COVID-19 led to the confinement of the entire population of Spain between March 15 and June 21, 2020. The subsequent evolution of the pandemic led to new partial confinements from October 25 of the same year until May 9, 2021. Consequently, on-site classes in educational centers were interrupted and were not resumed until September 2021 with limitations.

This situation limited the coexistence of educational communities and forced virtual teaching through different media. In parallel, the use of social networks was increased as a form of remote communication (INE, 2021), singularly on Twitter which is the social network preferred by teaching communities in Spain (Marcelo and Marcelo, 2021; Santoveña-Casal and Bernal-Bravo 2019).

Globally, Twitter picked up a significant increase in negative sentiment in the pandemic period (Lwin et al., 2020). In the case of teachers, studies in other countries showed that positive and negative affective postings increased during COVID-19 (Alwafi, 2021), although the main content dealt with Information and Communication Technologies (ICTs) use. In Spain, the confinement had emotional consequences on teachers and caused teachers to alter their emotional state, which was reflected in an increase in symptoms of stress, anxiety and depression (Ozamiz-Etxebarria, 2021).

This paper presents an exploratory study on the emotional content and topics discussed by Twitter teacher communities in Spain during the confinement period, before and after the confinement. Several studies support the use of educational hashtags as a means to organize teacher conversations and analyze their contents (Greenhalgh et al., 2020; Greenhow et al., 2021; Willet, 2019). Consequently, in this study we turned to the most used educational hashtags in Spain between 2018 and 2022, among which #claustrovirtual (#virtualteachingstaff) stood out.

Literature review

Educational communities on Twitter

Until the creation and rise of social networks, one of the main obstacles faced by teachers was the creation of communication networks that would allow them to connect with colleagues from different educational centers. Today, this issue has been largely overcome with the popularization of social networks, which expand the possibilities of interaction with other teachers (Carpenter et al., 2020).

Since the birth of Twitter, this social network was used by teachers to collaborate, share experiences, train and establish new professional contacts (Carpenter and Krutka, 2015; Greenhow et al., 2021; Krutka and Carpenter, 2016).

Although it is very common for teachers to use Twitter to carry out activities with their students (Carpenter et al., 2020; Nochumson, 2020; Rosenberg et al., 2016; Xing and Gao, 2018), it is also a network widely used by teachers to keep in touch with other teachers with common interests (Luo et al., 2020).

Teachers share their experiences on Twitter and explore new ideas especially on the educational use of ICTs and professional development (Willet, 2019). Educational resources are also shared (Carpenter and Krutka, 2014; Rehm and Notten, 2016).

In addition to this content, informal online communities and networks of teachers offer emotional support (Luo and Clifton, 2017; Macià and García, 2016; Noble et al., 2016) and reinforce identity as teachers (Carpenter et al., 2020).

Literature reviews by Galvin and Greenhow (2020) and Macià and García (2016) showed that in teachers’ use of social networks, professional learning prevailed over classroom instruction. Specifically, they were used to ask and answer questions; share and find teaching-related resources; reflect; dialogue; and obtain emotional support.

With respect to the evolution of content addressed by teachers, prior to the pandemic teachers used social networks to find instructional supports and for professional development (Bruguera et al., 2019). However, during the pandemic social networks were also used to seek help with the immediate challenge of school closures and emergency online instruction (Greenhow et al., 2021).

Sentiment analysis

Sentiment Analysis (SA) also called Opinion Mining (OM) is the field of study that analyzes people’s opinions by assessing the feelings, attitudes, and emotions that are generated towards issues such as products, services, issues, or topics, among others (Liu, 2012). Although two levels of SA are usually distinguished, one for the whole document and another for each of the sentences (Liu, 2012; Medhat et al., 2014), in the case of Twitter, the analysis of each tweet allows the unification of both analyses due to the limitation of characters and the brevity of these texts.

The systematic review of the Spanish language-specific literature on this topic by Osorio et al. (2021) is particularly useful for this study given its specific focus on the Spanish language. Although Spanish is the third most used language on the Internet after English and Chinese, SA resources for this language are scarce.

Machine Learning-based SA is the favorite for Spanish language SA. In contrast, Lexicon-based SA (LB) is the least used in Spanish language and is usually translated from English (Osorio et al., 2021; Saura et al., 2018).

SA from emojis is increasingly used on Twitter because emojis act as strong indicators of sentiment and are clearly identified when analyzed from a Natural Language Processing (NLP) perspective (Novak et al., 2015; Oyinbo and Olaniyi, 2021; Rathan et al., 2018). The study by Novak et al. (2015) identifies positive emotional content in most emojis, especially the most popular ones (). The research by Fernandez-Gavilanes et al. (2018) delves into SA in emojis also in the context of the Spanish language and concludes that emoticons have proven to be useful in sentiment classification.

In the educational context, SA was an under-exploited resource (Chauhan et al., 2019), although in recent years it has experienced a remarkable increase (Baragash and Aldowah, 2021). Koehler et al. (2015) discuss the potential applications of SA in educational research and practice, especially on Twitter, and highlight features related to online teaching, professional development and educational communities (e.g. monitor students’ online conversations in real time, learn about students’ motivation and understanding of materia, and measure the “course mood”).

Mite-Baidal et al. (2018) and Zhou and Ye (2020) conduct a systematic review of the literature on SA in education and conclude that some of the main benefits of using SA in the educational setting are the improvement of the teaching-learning process and student achievement, as well as the reduction of course dropout. None of these studies refer to SA of teacher networks.

Aim and research questions

In this study, we conducted an exploratory study about the interests and feelings of teachers on Twitter and how these have evolved in the last 4 years due to the exceptional circumstances that have been experienced worldwide. For this purpose, this study has differentiated between three critical periods to establish the chronological evolution of teachers’ opinions and interests. Specifically, the study has been divided into pre-pandemic, confinement and state of alarm, and post-pandemic periods.

The analysis of these three distinctive periods allows us to advance the knowledge base on the role played by the social network Twitter in the educational response of teachers to the COVID-19 pandemic. To this end, the present study was guided by the following research questions:

- RQ1. What was the overall activity of teachers on Twitter before, during, and after the pandemic?

- RQ2. How did the polarity of sentiment evolve in teachers’ Twitter posts before, during, and after the pandemic?

- RQ3. What were the main topics for teachers on Twitter before, during, and after the pandemic?

Methodology

A mixed methodology based on public data mining, semantic content analysis, and SA was used. This methodology involves the use of digital crawling data. It aims to more efficiently collect, organise and analyse generalisable samples of data representing people in virtual learning (Kimmons and Veletsianos, 2018).

Sample

The sample was extracted through the Twitter v2 API search tool. The hashtags #eduhora (#eduhour), #claustrovirtual (#virtualteachingstaff), #SerProfeMola (#beingateacheriscool), #otraeducaciónesposible (#anothereducationisposible), #claustrotuitero (#twitterteachingstaff), #profesquemolan (#coolteachers), #orgullodocente (#teacherpride), and #soymaestro (# Iamateacher) were identified. The requests were made on a monthly basis between January 2018 and March 2022. This process resulted in the identification of 30,752 tweets, from which were extracted those that not only had textual content, but also emotional and/or visual content reflected in the use of emojis. The final sample was reduced to 9462 tweets.

Teachers’ interventions on Twitter were analyzed according to the chronological evolution of the pandemic in Spain: from January 2018 to February 2020 (pre-pandemic period); from March 2020 to May 2021 (pandemic period); and from June 2021 to March 2022 (post-pandemic period), respectively.

The final dataset within each specified time frame included 3102 tweets for the prepandemic period (32.8%), 3728 tweets for the pandemic period (39.4%), and 2632 tweets for the post-pandemic period (27.8%).

Data analysis

For the analysis of the tweets, punctuation marks and unnecessary characters were removed from the JavaScript Object Notation (JSON) encoding of the Twitter API.

To answer the first research question, the number of tweets and the number of emojis in the sample were compared by month.

To answer the second research question, SA follows the model proposed by De Albornoz et al. (2012). First, the emotional content of each tweet is identified to different degrees. Second, tweets are classified according to sentiment polarity (positive or negative). Finally, Finally, SA is completed with visual data extracted from emojis. The results of these analyses are triangulated to determine the emotional charge of the analyzed tweets from the NLP point of view. To determine the attitude or polarity of the analyzed opinions, other previous research has been considered (Carpenter et al., 2020; Guerris et al., 2020; Kimmons et al., 2018; Xing and Gao, 2018).

The emotional content of each tweet was scored with Google’s Natural Language API that performs sentiment analysis, providing information on a scale with a numeric sentiment value ranging from −1 to +1. The Google API score was converted into a text label taking as a reference the formula validated by Quintana-Gomez (2021). Thus, values collected between 1 and 0.3 were considered positive, between 0.2 and −0.2 neutral, and between −0.3 and −1 negative.

The classification of tweets according to sentiment polarity (positive or negative) was also performed based on a semantic analysis of LB (Moreno-Ortiz, 2017; Peña et al., 2018). For this analysis, we chose to use the lexicon proposed by Hu and Liu (2004) translated from English to Spanish (Osorio et al., 2021).

Finally, SA was performed from visual data identifying the most frequent emojis in the sample (Fernández-Gavilanes et al., 2018; Novak et al., 2015; Oyinbo and Olaniyi, 2021). The occurrence of these emojis was related to previous SA results and to the polarity of sentiment in the analyzed tweets.

Results

RQ1. What was the overall activity of teachers on Twitter before, during and after the pandemic?

Figure 1 shows the total number of interventions generated on Twitter in the educational communities analyzed before, during, and after COVID-19. Overall, the number of Twitter interventions that included emojis increased significantly over the entire period studied. The first relevant peak occurring is generated during the pandemic in June 2020. After the pandemic, the number of interventions declines until October 2021, when it increases again and the maximum number of tweets within the analyzed period is reached. This increase coincides with the return to the classroom.

RQ2. How did the polarity of sentiment evolve in faculty interventions on Twitter before, during and after the pandemic?

Figure 1.

Tweets and emojis before, during and after COVID-19.

SA with Google’s Natural Language API shows a clearly positive polarity in tweets posted by faculty taking the entire sample as a whole. 69% (n = 6525) of the tweets analyzed presented a positive score between +0.3 and +1. This is followed by tweets with neutral emotional content, representing 29.5% (n = 2793). Tweets in which negative sentiments are identified were the least frequent with 1.5% (n = 144) with respect to the total sample (Figure 2).

Figure 2.

SA of total tweets.

These data were then broken down into the three time periods studied to identify the trend in each of them. As shown in Table 1, during the pandemic period, the emotional content of the tweets increases. Although positive sentiments continue to be the most frequent, an increase in negative and neutral sentiments is also observed. This trend is not observed in either the pre-pandemic or post-pandemic period.

Table 1.

Frequency and percentages extracted from ES for each period.

Study period	Negative		Neutral		Positive
Study period	f	%	f	%	f	%
Before Covid-19	36	25	829	29.7	2237	34.3
During Covid-19	69	47.9	1139	40.8	2520	38.6
After Covid-19	39	27.1	825	29.5	1768	27.1
Total	144	100	2793	100	6525	100

In general, tweets posted by teachers are characterized by the presence of positive sentiments, and the scarcity of tweets with negative sentiments. In the pandemic period in Spain, the percentage of tweets with positive emotional content increases significantly with respect to the previous period, reaching its maximum increase in two key moments: June 2020 and May 2021. This increase coincides with the end of the school periods of the 2019/2020 and 2020/2021 courses, which at those times were in virtual mode without face-to-face classes. Tweets with negative emotional content also increase in the pandemic period. The percentage of positive opinions decreased slightly after the pandemic, although positive polarity continued to set the trend (Figure 3).

Figure 3.

Evolution of sentiment polarity in teacher interventions on Twitter before, during and after the COVID-19 pandemic in Spain.

Semantic analysis was performed with SE, which extracted 3561 nouns, 1179 adjectives, and 1371 verbs from the sample. The localized terms were crossed with the Spanish translation of the lexicon proposed by Hu and Liu (2004) for LB. The result provided 94.6% of positive terms and 5.4% of negative terms. These percentages coincide with the polarity marked by the SA based on machine learning with Google’s Natural Language API. The positive terms that appeared most frequently in the teacher’s tweets in Spain were “improve”, “help”, “delighted” and “recommend”. As for negative terms, the most frequent were “struggle”, “crazy”, “worry”, “strange” and “shameless”.

The evolution of the use of these terms in the three periods studied offers interesting results. For example, the term “help/help” in the pre-pandemic period is the fifth most frequent term in the corpus with a frequency of 82, while in the pandemic period it is in first position with a frequency of 248. In the post-pandemic period it drops to second place with a frequency of 106. The same is true for the negative terms. Table 2 presents the complete analysis for the three periods studied.

Table 2.

Frequency extracted from the LB for each of the periods studied.

	Before Covid-19		During Covid-19		After Covid-19
Positive
	f	%	f	%	f	%
Better/improve	187	23.7	199	19.2	177	23.3
Creative	39	11.9	75	7.2	69	9.1
Enchanted/enchant	39	4.9	99	9.6	50	6.6
Encourage	56	7.1	84	8.1	54	7.1
Enjoy	85	10.8	68	6.6	70	9.2
Great	83	10.5	76	7.3	85	11.2
Help	82	10.4	248	24.0	106	13.9
Inspire/inspiration	37	4.7	39	3.8	43	5.7
Recommend	65	8.2	96	9.3	58	7.6
Thankful/grateful	60	7.6	51	4.9	48	6.3
Negative
Crazy	22	39.3	40	51.9	21	51.2
Fight	26	46.4	0	0	6	14.6
Scoundrel	1	1.8	1	1.3	0	0
Stranger	1	1.8	5	6.5	2	4.9
Struggle	0	0	19	24.7	0	0
Worry	6	10.7	12	15.6	12	29.3

SA from visual data was based on the appearance of emojis with positive, neutral and negative emotional content, to contrast it with the results of SA based on machine learning. A total of 1089 different emojis were extracted, of which 910 were simple and 179 were composite. The emojis from the sample total whose frequency was at least 100 times were analyzed (Table 3).

Table 3.

Distribution of emojis by period and sentiment polarity.

Human face emoticons representing positive emotions predominated. For example, smiley face with smiling eyes, smiley face with heart eyes, smiley face with hearts, rolling on the floor laughing, hugging face, face blowing kiss, surprised by a star, smiley face with smiling eyes, winking face, smiley face with big eyes.

Comparison of the machine learning-based SA results (Table 1) and the total emoji percentages for each period (Table 3) show a very similar sentiment polarity. The results of SA from visual data based on emojis share a strong correlation with SA based on machine learning, which reinforces the idea of using them complementarily in this type of analysis as a sentiment label. Although the LB results divide sentiments into positive and negative (Table 2), the polarity of sentiments considering only these two categories also matches the SA results from visual data.

RQ3. What were the top topics for faculty on Twitter before, during, and after the pandemic?

As Zhou and Mou (2022, p. 11) note, “there is no universal rule for determining the number of existing topics in a dataset.” SE located in the tweets of each period the most frequent lexical combinations that constituted the topics of the sample (Table 5). For each of these topics, the SE keyness score value was extracted, which determines the relevance in the corpus of each period compared to the Spanish Web 2018 reference corpus of 16 billion words (Firoozeh et al., 2020). As a result, one topic was identified as relevant for the pre-pandemic period, four topics for the pandemic period and two topics for the post-pandemic period. (Table 4).

Table 4.

Weight in the Topics ID corpus based on keyness Score (SE).

Topic ID	Before Covid-19	During Covid-19	After Covid-19
Keyness score
1	185,8	444,5	263,2
2	52,9	334,8	255,2
3	35,3	274,6	79,8
4	30,8	246,8	71,8
5	18,7	141,2	55,8

Table 5 shows the relevance of the topics in the pre-pandemic, pandemic and post-pandemic periods. Before the pandemic, the preferred topic for teachers was mainly related to digital tools and didactic methodologies such as gamification.

Table 5.

Topics generated by SE analysis.

Topic	Terms under each topic
Before Covid-19
1	Genially; realidad aumentada; realidad virtual; gamificación; canvas (genially; augmented reality; virtual reality; gamification; canvas)
During Covid-19
1	Yo me quedo en casa; quédate en casa; docentes en casa; superhéroes; confinamiento; cuarentena; coronavirus; mascarilla; yo me vacuno; nueva normalidad; desescalada (I stay at home; stay at home; teachers at home; superheroes; confinement; quarantine; coronavirus; mask; I get vaccinated; new normal; de-escalation)
2	Educación a distancia; semipresencial; clases on line; clases virtuales; aula virtual; videollamada; webinar; webinally; aprendizaje a distancia; educación en línea (distance education; blended learning; online classes; virtual classes; online education; virtual classroom; videoconferencing; webinar; webinally; distance learning; distance learners; online education)
3	Vuelta segura; vuelta al cole segura; vuelta al cole; vuelta segura a las aulas; vuelta sin riesgo (safe back home; safe back to the classroom; safe back to school; safe back to school; risk-free return)
4	Genially; realidad aumentada; realidad virtual; gamificar; gamificación; canvas; padlet (genially; augmented reality; virtual reality; gamify; gamification; canvas; padlet)
After Covid-19
1	Genially; realidad aumentada; realidad virtual; gamificar; gamificación; canvas; padlet (genially; augmented reality; virtual reality; gamify; gamification; canvas; padlet)
2	Videollamada; webinar; educación en línea; video clase; videoconferencia; aprende en casa; el cole en casa (video call; webinar; online education; video class; videoconference; learn at home; school at home)

During the pandemic, four relevant topics were identified on Twitter for teachers. The first topic focused on the situation of confinement that was being experienced in Spain, the health measures and the transition to the new normality. Topic 2 was related to issues related to virtual teaching. Topic 3 reflected teachers’ concerns about the safety of the return to the classroom. Finally, topic 4 dealt with digital tools and teaching strategies, and is identified with the topic of the pre-pandemic period.

After the pandemic, the topics are reduced. Topic 1 deals again with digital tools and didactic strategies; this topic is repeated in all the analyzed periods. In this period after the confinement, a second topic appears, which is related to the first one and refers to distance communication.

Discussion

This study aims to analyze the topics of interest and sentiments of teachers in Spain before, during and after COVID-19. Interventions on Twitter over the last 4 years have been examined, which has allowed us to analyze whether significant changes have occurred in three key periods. On the one hand, a pre-pandemic period, on the other the pandemic moment linked in Spain to the confinement of citizens and the state of alarm, and finally the post-pandemic period beginning with the end of the state of alarm.

To answer the first research question on general activity, the results show that teacher use of the social network Twitter increases from 2018 to 2022. As we enter the pandemic phase, the number of interventions increased very significantly, with the high point being in June 2020, with the end of the school year that in March had become virtually taught for all school stages. These results coincide with the studies of Greenhow et al. (2021) and Alwafi (2021) for other countries with a similar situation of confinement and virtual teaching. This result is also in line with other researches that show that emergency situations influence the concern of citizens in general, and of the educational community in particular, for those issues that affect the educational environment (Li et al., 2021; Trust et al., 2020; Zhou and Mou, 2022).

With respect to the second research question, SA highlights that teachers have maintained a generally positive view, even during the pandemic, which has been contrasted with previous studies (Chanchí et al., 2021). The findings of this study further show that both tweets linked to positive sentiments and those linked to negative sentiments increased during COVID-19. As the results of Arora et al. (2021) and those of Alwafi (2021) had already established, the number of tweets linked to positive sentiments was higher than the number of tweets linked to negative sentiments.

The emoji-based SA results showed a clearly positive polarity that confirmed the results of other sentiment analysis methods, which supports the use of SA from visual data. These results are in agreement with those of Novak et al. (2015) and Fernandez-Gavilanes et al. (2018) also for the Spanish language.

In response to the third research question, the keywords extracted from the corpus provide a detailed description of the issues that were of interest to Spanish teachers before, during and after the pandemic. Interest in digital tools in education and innovative methodologies remained constant in the three periods analyzed, although with different relevance. However, in the pandemic period new topics related to the implications of confinement and the evolution of COVID-19 appeared. These data reinforce the conclusions of previous studies that emphasize the importance of Twitter for creating collaborative support networks among teachers (Alwafi, 2021; Carpenter et al., 2020; Carpenter and Krutka, 2014, 2015).

Although there have been previous studies on teacher interaction on Twitter, our data shows for the first time the topics of interest and feelings of Spanish teachers on Twitter at three key moments. This research broadens the field of knowledge and allows comparative studies to be carried out to highlight similarities and differences with other countries.

Limitations of the study

Among the limitations of this study, we can point out that it is limited to teachers in Spain. Future research could be expanded to include teachers from different Spanish-speaking countries, resulting in a comparative study with a much larger sample. Another limitation of the study is that although a mixed methodology based on public data mining, semantic content analysis, and SA was used to examine teachers’ professional networks, most of the data are quantitative. It would be valuable to use other qualitative methods (e.g., interviews or focus groups) to obtain more detailed information about faculty concerns, interests, and feelings.

Conclusions and implications

The global pandemic resulting from Covid-19 has forced the educational community to operate in a digital environment, which has resulted in an increase in the use of Twitter as a social network for teachers in Spain. The extreme situation of confinement corresponds to the maximum activity of teachers to maintain active digital educational communities.

The emotional content of teachers’ tweets presents a clearly positive polarity before, during and after the pandemic. However, in the period of confinement the emotional content of the messages increases very significantly, as well as the use of emojis. This increase implies that teachers found the ideal forum to express both positive and negative feelings in the educational Twitter communities during the confinement. The results of sentiment analysis with machine learning, LB and visual data show the same polarity and consistent percentages, which endorses the latest sentiment analysis method as a valid complement in the analysis of emotional content on Twitter. Digital tools and innovative methodologies are the preferred topics for Spanish teachers. However, during the confinement, the preferred topics were the health alert and the conditions of the return to work in the classrooms, which were perceived with uncertainty.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Alejandro Gomez-Camacho

Author Biographies

Alejandro Gómez-Camacho is Associated Professor in the Department of Language Education at the University of Seville. He focuses his research on pedagogical ideas in the Spanish Golden Age, textisms and Spanish digital norm and Spanish language teaching.

Olga Moreno-Fernández is Associate Professor in the Department of Didactics of Experimental and Social Sciences in the Faculty of Education Sciences at the University of Seville. She is currently researching on the use of social networks by the educational community in Spain.

References

Alwafi

(2021) Tracing changes in teachers' professional learning network on Twitter: comparison of teachers' social network structure and content of interaction before and during the COVID-19 pandemic. Journal of Computer Assistant Learning 37: 1653–1665. DOI: 10.1111/jcal.12607.

Arora

Chakraborty

Bhatia

MPS

, et al. (2021) Role of emotion in excessive use of twitter during COVID-19 imposed lockdown in India. Journal of Technology in Behavioral Science 6: 370–377. DOI: 10.1007/s41347-020-00174-3.

Baragash

Aldowah

(2021) Sentiment analysis in higher education: a systematic mapping review. Journal of Physics: Conference Series 1860(1): 012002.

Bruguera

Guitert

Romeu

(2019) Social media and professional development: a systematic review. Research in Learning Technology 27: 1–18. DOI: 10.25304/rlt.v27.2286.

Carpenter

Krutka

(2014) How and why educators use twitter: a survey of the field. Journal of Research on Technology in Education 46(4): 414–434. DOI: 10.1080/15391523.2014.925701.

Carpenter

Krutka

(2015) Engagement through microblogging: educator professional development via Twitter. Professional Development in Education 41(4): 707–728. DOI: 10.1080/19415257.2014.939294.

Carpenter

Tani

Morrison

, et al. (2020) Exploring the landscape of educator professional activity on Twitter: an analysis of 16 education-related Twitter hashtags. Professional Development in Education 48(5): 784–805. DOI: 10.1080/19415257.2020.1752287.

Chanchí

Ospino

García

MÁ

(2021) Emotions and feelings in university teachers during the COVID-19 pandemic. Revista Venezolana de Gerencia 26(5): 123–139. DOI: 10.52080/rvgluz.26.e5.9.

Chauhan

Agrawal

Meena

(2019) Aspect-based sentiment analysis of students’ feedback to improve teaching-learning process. In: Satapathy

Joshi

(eds) Information and Communication Technology for Intelligent Systems. Smart Innovation, Systems and Technologies. Berlin: Springer, 259–266. DOI: 10.1007/978-981-13-1747-7_25.

10.

De Albornoz

Plaza

Gervas

(2012) SentiSense: an easily scalable concept-based affective lexicon for sentiment analysis Proceedings of the Eighth International Conference on Language Resources and Evaluation, Istanbul, Turkey, May, 2012, 3562–3567.

11.

Fernández-Gavilanes

Juncal-Martínez

García-Méndez

, et al. (2018) Creating emoji lexica from unsupervised sentiment analysis of their descriptions. Expert Systems with Applications 103(1): 74–91. DOI: 10.1016/j.eswa.2018.02.043.

12.

Firoozeh

Nazarenko

Alizon

, et al. (2020) Keyword extraction: issues and methods. Natural Language Engineering 26(3): 259–291. DOI: 10.1017/S1351324919000457.

13.

Galvin

Greenhow

(2020) Educational networking: a novel discipline for improved K-12 learning based on social networks. In: Peña-Ayala

(ed) Educational Networking: A Novel Discipline for Improved Learning Based on Social Networks. Berlin: Springer, 3–41.

14.

Greenhalgh

Rosenberg

Willet

KBS

, et al. (2020) Identifying multiple learning spaces within a single teacher-focused Twitter hashtag. Computers & Education 148: 103809. DOI: 10.1016/j.compedu.2020.103809.

15.

Greenhalgh

Nnagboro

Kaufmann

, et al. (2021) Academic, social, and cultural learning in the French #bac2018 Twitter hashtag. Educational Technology Research & Development 69: 1835–1851. DOI: 10.1007/s11423-021-10015-6.

16.

Greenhow

Willet

KBS

Galvin

(2021) Inquiring tweets want to know: #Edchat supports for #RemoteTeaching during COVID-19. British Journal of Educational Technology 52: 1434–1454. DOI: 10.1111/bjet.13097.

17.

Guerris

Cuadros

González-Sabaté

, et al. (2020) Describing the public perception of chemistry on twitter. Chemistry Education: Research and Practice 21(3): 989–999. DOI: 10.1039/C9RP00282K.

18.

Liu

(2004) Mining and summarizing customer reviews Proceedings of the ACM SIGKDD. International Conference on Knowledge Discovery and Data Mining, 22 August 2004. DOI: 10.1145/1014052.1014073.

19.

INE. Instituto Nacional de Estadística [National Statistical Institute] (2021) Población que usa Internet (en los últimos tres meses). Tipo de actividades realizadas por Internet [Population using the Internet (in the last three months). https://www.ine.es/ss/Satellite?L=es_ES&c=INESeccion_C&cid=1259925528782&p=1254735110672&pagename=ProductosYServicios%2FPYSLayout#:∼:text=En_el_a%C3%B1o_2021_en,33%2C1_millones_de_usuarios.

20.

Kimmons

Veletsianos

(2018) Public internet data mining methods in instructional design, educational technology, and online learning research. TechTrends 62: 492–500. DOI: 10.1007/s11528-018-0307-4.

21.

Kimmons

Carpenter

Veletsianos

, et al. (2018) Mining social media divides: an analysis of K-12 U.S. School uses of Twitter. Learning, Media and Technology 43(3): 307–325. DOI: 10.1080/17439884.2018.1504791.

22.

Koehler

Greenhalgh

Zellner

(2015) Potential applications of sentiment analysis in educational research and practice - is SITE the friendliest conference? In: Rutledge

Slykhuis

(eds) Proceedings of SITE 2015--Society for Information Technology & Teacher Education International Conference. Waynesville, NC, Mar 02, 2015, Association for the Advancement of Computing in Education (AACE), 1348–1354.

23.

Krutka

Carpenter

(2016) Participatory learning through social media: how and why social studies educators use Twitter. Contemporary Issues in Technology and Teacher Education 16(1): 38–59.

24.

Huo

Lin

JCW

(2021) Cross-dimension mining model of public opinion data in online education based on fuzzy association rules. Mobile Networks and Applications (26): 2127–2140. DOI: 10.1007/s11036-021-01769-7.

25.

Liu

(2012) “Sentiment analysis and opinion mining”, synthesis lectures on human language. Technologies 5(1): 1–167.

26.

Luo

Clifton

(2017) Examining collaborative knowledge construction in microblogging-based learning environments. Journal of Information Technology Education: Research 16(16): 365–390.

27.

Luo

Freeman

Stefaniak

(2020) “Like, comment, and share” professional development through social media in higher education: a systematic review. Educational Technology Research & Development 68(4): 1659–1683. DOI: 10.1007/s11423-020-09790-5.

28.

Lwin

Sheldenkar

, et al. (2020) Global sentiments surrounding the COVID-19 pandemic on twitter: analysis of twitter trends. JMIR Public Health Surveillance 6(2): e19447. DOI: 10.2196/19447.

29.

Macià

García

(2016) Informal online communities and networks as a source of teacher professional development: a review. Teaching and Teacher Education 55: 291–307. DOI: 10.1016/j.tate.2016.01.021.

30.

Marcelo

(2021) Educational influencers on Twitter. Analysis of hashtags and relationship structure. Comunicar 68: 73–83. DOI: 10.3916/C68-2021-06.

31.

Medhat

Hassan

Korashy

(2014) Sentiment analysis algorithms and applications: a survey. Ain Shams Engineering Journal 5(4): 1093–1113. DOI: 10.1016/j.asej.2014.04.011.

32.

Mite-Baidal

Delgado-Vera

Solís-Avilés

, et al. (2018) Sentiment analysis in education domain: a systematic literature review. In: Valencia-García

Alcaraz-Mármol

Del Cioppo-Morstadt

, et al. (eds) Technologies and Innovation. CITI 2018. Communications in Computer and Information Science. Berlin: Springer, 285–297. DOI: 10.1007/978-3-030-00940-3_21.

33.

Moreno-Ortiz

(2017) Lingmotif: a user-focused sentiment analysis tool. Procesamiento del Lenguaje Natural 58: 133–140.

34.

Noble

McQuillan

Littenberg-Tobias

(2016) “A lifelong classroom”: social studies educators’ engagement with professional learning networks on twitter. Journal of Technology and Teacher Education 24(2): 187–213.

35.

Nochumson

(2020) Elementary schoolteachers’ use of twitter: exploring the implications of learning through online social media. Professional Development in Education 46(2): 306–323. DOI: 10.1080/19415257.2019.1585382.

36.

Novak

Smailovic

Sluban

, et al. (2015) Sentiment of emojis. PLoS One 10(12): 1–22. DOI: 10.1371/journal.pone.0144296.

37.

Osorio

Peña

Espinoza-Valdez

(2021) Systematic literature review of sentiment analysis in the Spanish language. Data Technologies and Applications 55(4): 461–479. DOI: 10.1108/DTA-09-2020-0200.

38.

Oyinbo

Olaniyi

(2021) A multimodal analysis of emojis in selected whatsapp chats in social media discourse: a discourse analyst approach. Zamfara International Journal of Humanities (ZIJOH) 1(1-2): 110–120.

39.

Ozamiz-Etxebarria

Dosil

Mondragon

, et al. (2021) Emotional state of school and university teachers in northern Spain in the face of COVID-19. Revista Española de Salud Pública 95: 1–18.

40.

Peña

Casillas

Lara

, et al. (2018) Mexican Spanish affective Dictionary. In: Rocha

Adeli

Reis

, et al. (eds) Trends and Advances in Information Systems and Technologies. Berlin: Springer, 63–71. DOI: 10.1007/978-3-319-77703-0_6.

41.

Quintana-Gómez

(2021) Análisis de los procesos de tratamiento de información en un estudio de análisis de sentimiento utilizando la tecnología de Google. Vivat Academia. Revista de Comunicación 154: 41–55. DOI: 10.15178/va.2021.154.e1336.

42.

Rathan

Hulipalled

Venugopal

, et al. (2018) Consumer insight mining: aspect based twitter opinion mining of mobile phone reviews. Applied Soft Computing 68: 765–773. DOI: 10.1016/j.asoc.2017.07.056.

43.

Rehm

Notten

(2016) Twitter as an informal learning space for teachers!? The role of social capital in Twitter conversations among teachers. Teaching and Teacher Education 60: 215–223. DOI: 10.1016/j.tate.2016.08.015.

44.

Rosenberg

Greenhalgh

Koehler

, et al. (2016) An investigation of state educational twitter hashtags (SETHs) as affinity spaces. E-learning and Digital Media 13: 24–44. DOI: 10.1177/2042753016672351.

45.

Santoveña-Casal

Bernal-Bravo

(2019) Exploring the influence of the teacher: social participation on Twitter and academic perception. Comunicar 27: 75–84. DOI: 10.3916/c58-2019-07.

46.

Saura

Palos-Sanchez

Rios-Martin

(2018) Attitudes to environmental factors in the tourism sector expressed in online comments: an exploratory study. International Journal of Environmental Research and Public Health 15(3): 553. DOI: 10.3390/ijerph15030553.

47.

Trust

Carpenter

Krutka

, et al. (2020) #RemoteTeaching & #RemoteLearning: educator tweeting during the COVID-19 pandemic. Journal of Technology and Teacher Education 28(2): 151–159.

48.

Willet

KBS

(2019) Revisiting how and why educators use Twitter: tweet types and purposes in #Edchat. Journal of Research on Technology in Education 51: 273–289. DOI: 10.1080/15391523.2019.1611507.

49.

Xing

Gao

(2018) Exploring the relationship between online discourse and commitment in Twitter professional learning communities. Computers & Education 126: 388–398. DOI: 10.1080/19415257.2014.93929410.1016/j.compedu.2018.08.010.

50.

Zhou

Mou

(2022) Tracking public opinion about online education over COVID-19 in China. Educational Technology Research & Development 70(3): 1083–1104. DOI: 10.1007/s11423-022-10080-5.

51.

Zhou

(2020) Sentiment analysis in education research: a review of journal publications. Interactive Learning Environments 31(3). DOI: 10.1080/10494820.2020.1826985.