Yue SupersName, Jia XuepersName, Xiaoqian LiupersName, Peijing WupersName, Junxiang ChenpersName, Chen ChenpersName, Tianli LiupersName, Weigang GongpersName, Tingshao ZhupersName (2020date). “Examining the Impact of COVID-19 Lockdown in Wuhan and Lombardy: A Psycholinguistic Analysis on Weibo and Twitter.” International Journal of Environmental Research and Public Healthtitle·j.
Object of study — lockdown-associated changes in psycholinguistic featuresterm·dom of Weibo posts from Wuhan and Twitter posts from Lombardy
Research question — how psychological-state indicators in public social-media language changed before versus after lockdownterm·dom and differed across the two cultural settings
Contribution — a real-time, cross-cultural comparison using pre-lockdown digital traces that avoids relying solely on retrospective self-report
phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | lockdown coincided with shared increases in home and cognitive-process language but culturally divergent attention, emotion, uncertainty, and stress signals — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | |
| claim-type | existence — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | |
| method — evidence | — “Many countries are taking strict quarantine policies to prevent the rapid spread of COVID-19 (Corona Virus Disease 2019) around the world, such as city lockdown. Cities in China and Italy were locked down in the early stage of the pandemic. The present study aims to examine and compare the impact of COVID-19 lockdown on individuals’ psychological states in China and Italy. We achieved the aim by (1) sampling Weibo users (geo-location = Wuhan, China) and Twitter users (geo-location = Lombardy, Italy); (2) fetching all the users’ published posts two weeks before and after the lockdown in each region (e.g., the lockdown date of Wuhan was 23 January 2020); (3) extracting the psycholinguistic features of these posts using the Simplified Chinese and Italian version of Language Inquiry and Word Count (LIWC) dictionary; and (4) conducting Wilcoxon tests to examine the changes in the psycholinguistic characteristics of the posts before and after the lockdown in Wuhan and Lombardy, respectively. Results showed that individuals focused more on “home”, and expressed a higher level of cognitive process after a lockdown in both Wuhan and Lombardy. Meanwhile, the level of stress decreased, and the attention to leisure increased in Lombardy after the lockdown. The attention to group, religion, and emotions became more prevalent in Wuhan after the lockdown. Findings provide decision-makers timely evidence on public reactions and the impacts on psychological states in the COVID-19 context, and have implications for evidence-based mental health interventions in two countries.” | |
| object of study — evidence | — “Many countries are taking strict quarantine policies to prevent the rapid spread of COVID-19 (Corona Virus Disease 2019) around the world, such as city lockdown. Cities in China and Italy were locked down in the early stage of the pandemic. The present study aims to examine and compare the impact of COVID-19 lockdown on individuals’ psychological states in China and Italy. We achieved the aim by (1) sampling Weibo users (geo-location = Wuhan, China) and Twitter users (geo-location = Lombardy, Italy); (2) fetching all the users’ published posts two weeks before and after the lockdown in each region (e.g., the lockdown date of Wuhan was 23 January 2020); (3) extracting the psycholinguistic features of these posts using the Simplified Chinese and Italian version of Language Inquiry and Word Count (LIWC) dictionary; and (4) conducting Wilcoxon tests to examine the changes in the psycholinguistic characteristics of the posts before and after the lockdown in Wuhan and Lombardy, respectively. Results showed that individuals focused more on “home”, and expressed a higher level of cognitive process after a lockdown in both Wuhan and Lombardy. Meanwhile, the level of stress decreased, and the attention to leisure increased in Lombardy after the lockdown. The attention to group, religion, and emotions became more prevalent in Wuhan after the lockdown. Findings provide decision-makers timely evidence on public reactions and the impacts on psychological states in the COVID-19 context, and have implications for evidence-based mental health interventions in two countries.” | |
| research question — evidence | — “In this study, we used the simplified Chinese version of LIWC and Italian LIWC to measure people’s psychological status before and after the lockdown in Wuhan and Lombardy, respectively. By using psycholinguistic analysis, we aim to identify the psychological effects of the lockdown on individuals in Wuhan and Lombardy.” | |
| contribution — evidence | — “Recent studies used the self-report questionnaire approach to examine the psychological responses during the lockdown in different countries, including Italy, India, and China [ 12 , 16 , 17 ]. However, these studies rely on retrospective and time-lagged surveys and interviews. These approaches have limitations in accessing psychological statuses before the lockdown. That is, there is recall bias inevitably when people are required to recall a past period in the retrospective study.” | |
| mechanism | Mobility restrictions increased time at home and were reflected in greater home-word use. — “The increased use of home words is related to mobility control after the lockdown in Wuhan and Lombardy. Researchers estimated that mobility and social contacts in China during the lockdown dropped about 80%, concerning a baseline set on 1 January 2020 [ 30 ]. Moreover, Google reported a 23% increase in residential location activity in Lombardy during lockdown compared to baseline [ 31 ]. These reports indicate people spend more time at home and spend less time outdoors during a lockdown, which is consistent with more use of home words.” | mechanisms[0] |
| mechanism | Cognitive-process words were interpreted as attempts to make sense of changed circumstances. — “The frequency of discrepancy words increases after the lockdown in Wuhan and Lombardy. Besides, we observe the increased use of inhibition and certain words after the lockdown in Wuhan. Previous study suggests that the uses of discrepancy, Inhibition, and certain words reflect the change of degree of cognitive processing [ 32 ]. Furthermore, cognitive processing indicates that individuals make efforts to make sense of the environment [ 32 ]. Residents in Wuhan and Lombardy attempt to figure out what has happened after the lockdown. Thus, they could adjust their attitudes and lifestyles to accommodate new circumstances during the COVID-19 pandemic.” | mechanisms[1] |
| mechanism | Reduced first-person singular and increased plural pronouns marked attention shifting toward communities. — “In Wuhan, the uses of the first-person plural pronouns, second-person plural pronoun increase after a lockdown, while the use of first-person singular pronoun decreases. Previous reports confirm that first-person singular pronouns show attention to the self, whereas most other pronouns suggest attention to other individuals [ 36 ]. Moreover, “we” implies a sense of group identity sometimes [ 37 ]. Results suggest that people switch their attention from themselves to others and the communities after the lockdown. Besides, the increased use of “we” indicates that people focus more on the group, become more united, and share more group identity after a lockdown, which is consistent with some researchers’ opinions [ 1 ]. China has a collectivistic culture, and Italy has an individualistic culture [ 38 ]. Results show that the increased use of other pronouns and decreased use of first-person singular pronouns suggest a collectivistic culture in China. At the same time, the absence of such a consequence in Lombardy might be related to the individualist culture. Researchers find that people sharing collectivist values stress more communal coping as a resource to cope with collective traumatic events [ 39 ], which is consistent with our research conducted in the context of the lockdown.” | mechanisms[2] |
| mechanism | Frequent emotion words indexed deeper immersion in negative events. — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | mechanisms[3] |
| mechanism | Social-word use marked attention to coping resources and social support. — “The increased use of social words in Wuhan after the lockdown suggests the focus on social concerns and social support [ 22 ]. Social support can make people feel better about their situation and reinforce the belief that they have access to support resources [ 16 , 42 ]. Thus, seeking social support is considered adaptive for people during a lockdown. In contrast, we do not observe such a change in Lombardy.” | mechanisms[4] |
| mechanism | Platform-specific posting conventions could create apparent cultural differences. — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | mechanisms[5] |
| sense | representation - normalized word-category frequencies standing in for otherwise unobserved psychological status — “The present study uses the Chinese version of LIWC and Italian version of LIWC to extract the psycholinguistic features from social media users’ posts. Examinations of the features allow us to access the changes of psychological status before and after the lockdown in Wuhan and Lombardy.” | senses[0] |
| sense | attention - the social target of language inferred from pronoun and content-word frequencies — “In Wuhan, the uses of the first-person plural pronouns, second-person plural pronoun increase after a lockdown, while the use of first-person singular pronoun decreases. Previous reports confirm that first-person singular pronouns show attention to the self, whereas most other pronouns suggest attention to other individuals [ 36 ]. Moreover, “we” implies a sense of group identity sometimes [ 37 ]. Results suggest that people switch their attention from themselves to others and the communities after the lockdown. Besides, the increased use of “we” indicates that people focus more on the group, become more united, and share more group identity after a lockdown, which is consistent with some researchers’ opinions [ 1 ]. China has a collectivistic culture, and Italy has an individualistic culture [ 38 ]. Results show that the increased use of other pronouns and decreased use of first-person singular pronouns suggest a collectivistic culture in China. At the same time, the absence of such a consequence in Lombardy might be related to the individualist culture. Researchers find that people sharing collectivist values stress more communal coping as a resource to cope with collective traumatic events [ 39 ], which is consistent with our research conducted in the context of the lockdown.” | senses[1] |
| sense | emotion - public use of affect and negative-emotion word categories — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | senses[2] |
| sense | control - both imposed mobility restriction and perceived capacity to manage one's life — “Study findings have implications for decision-makers, public health authorities, and practitioners. First, considering the efforts of adjusting to the changing environment in both Wuhan and Lombardy after the lockdown, decision-makers should ensure the supply chain functions as usual to ensure people’s confidence in having the control of their lives. Besides, public health authorities and practitioners could adjust their focus of service given the changes in residents’ attention after lockdown. For example, people in Wuhan expressed more stress and negative emotions, public health authorities and practitioners should take interventions to comfort them and relieve stress, such as the online consulting service and indoor activities. Notably, the support for individuals with pre-existing mental or physical health issues is also needed. Meanwhile, people did not show significant stress in Lombardy. Public health communities and practitioners might focus more on the popularization of pandemic prevention knowledge and the reinforcement of protection awareness.” | senses[3] |
| sense | bias - systematic platform and language differences that threaten a cultural comparison — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | senses[4] |
| control | within-user pre-post grouping — held — “We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdown” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdown” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWC dictionary containing 51 common word categories to extract psycholinguistic features and calculated word frequencies of each word category in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.” — rules out stable individual differences being mistaken for lockdown change — “We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdown” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdown” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWC dictionary containing 51 common word categories to extract psycholinguistic features and calculated word frequencies of each word category in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.” | controls[0] |
| control | shared-category harmonization — partial — “We extracted the linguistic features using the Simplified Chinese LIWC dictionary (SCLIWC) [ 27 ] and the Italian LIWC dictionary [ 25 ]. Given both SCLIWC and Italian LIWC share LIWC dictionary structure, there are many common words in SCLIWC and LIWC. To make the result of Wuhan and Lombardy comparable, we only analyzed the common word categories between SCLIWC and Italian LIWC. The selection procedure of common categories is as follows:” — rules out language-specific dictionary categories driving cross-region differences — “We extracted the linguistic features using the Simplified Chinese LIWC dictionary (SCLIWC) [ 27 ] and the Italian LIWC dictionary [ 25 ]. Given both SCLIWC and Italian LIWC share LIWC dictionary structure, there are many common words in SCLIWC and LIWC. To make the result of Wuhan and Lombardy comparable, we only analyzed the common word categories between SCLIWC and Italian LIWC. The selection procedure of common categories is as follows:” | controls[1] |
| control | individual-account criterion — held — “The Chinese samples are from the Weibo data pool containing 1.16 million active Weibo users [ 28 ]. In this study, we selected active Weibo users from the data pool by the following criteria: Published at least one original post on average each day from 9 January 2020 to 5 February 2020 (i.e., two weeks before and after the lockdown); Individual users only, excluding any organizations; Locate at “Wuhan, Hubei” by the geo-location in the user profile.” — rules out organizational communication being mixed with personal posts — “The Chinese samples are from the Weibo data pool containing 1.16 million active Weibo users [ 28 ]. In this study, we selected active Weibo users from the data pool by the following criteria: Published at least one original post on average each day from 9 January 2020 to 5 February 2020 (i.e., two weeks before and after the lockdown); Individual users only, excluding any organizations; Locate at “Wuhan, Hubei” by the geo-location in the user profile.” | controls[2] |
| control | Italian-only tweet filter — held — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” — rules out non-Italian messages contaminating Italian LIWC counts — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” | controls[3] |
| control | region-specific date boundary — held — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” — rules out posts being assigned to the wrong lockdown period — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” | controls[4] |
| audit | lockdown —causes→ psycholinguistic feature — Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span compares features before and after lockdown; a pre-post change typed as cause, unsigned. — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | concept_edges[0] |
| relation | lockdown —causes→ psycholinguistic feature — The analysis attributes pre-post feature shifts to the lockdown event. — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | concept_edges[0] |
| weight | lockdown —causes→ psycholinguistic feature — 0.95 — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | concept_edges[0] |
| outcome | control — held — “We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdown” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdown” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWC dictionary containing 51 common word categories to extract psycholinguistic features and calculated word frequencies of each word category in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.” | controls[0] |
| outcome | control — partial — “We extracted the linguistic features using the Simplified Chinese LIWC dictionary (SCLIWC) [ 27 ] and the Italian LIWC dictionary [ 25 ]. Given both SCLIWC and Italian LIWC share LIWC dictionary structure, there are many common words in SCLIWC and LIWC. To make the result of Wuhan and Lombardy comparable, we only analyzed the common word categories between SCLIWC and Italian LIWC. The selection procedure of common categories is as follows:” | controls[1] |
| outcome | control — held — “The Chinese samples are from the Weibo data pool containing 1.16 million active Weibo users [ 28 ]. In this study, we selected active Weibo users from the data pool by the following criteria: Published at least one original post on average each day from 9 January 2020 to 5 February 2020 (i.e., two weeks before and after the lockdown); Individual users only, excluding any organizations; Locate at “Wuhan, Hubei” by the geo-location in the user profile.” | controls[2] |
| outcome | control — held — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” | controls[3] |
| outcome | control — held — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” | controls[4] |
| limitation | Selected active social-media users do not represent the full regional populations. — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | limitations[0] |
| limitation | Chinese-Italian language differences and apostrophe handling could introduce processing error. — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | limitations[1] |
| limitation | Locations were self-reported rather than verified from user IP addresses. — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | limitations[2] |
| limitation | Platform affordances and positivity differences could bias the cross-region contrast. — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | limitations[3] |
| limitation | Contrasts with survey findings may reflect different methods, designs, measures, and timeframes. — “The use of anxiety words decreases in Lombardy. Anxiety reveals self-reported stress [ 34 ]. Our results imply that people feel less stressed after Lombardy lockdown. However, people do not experience any change of stress in Wuhan. Researchers find that unrealistic optimism is more evident for European North Americans [ 13 ], which might be related to the different responses in the level of stress between Lombardy and Wuhan after the lockdown. However, our results are not consistent with existing studies [ 12 , 35 ]. Rossi and colleagues consider that the strict measures of the lockdown in Italy serve as an unprecedented stressful event [ 12 ]. Besides, Ahmed and colleagues find that 29% of respondents report different levels of anxiety related to lockdown at home in China [ 35 ]. Such differences could be due to different research methods, design, measurements, and timeframe used in the study.” | limitations[4] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | Home and discrepancy word frequencies increased in both Wuhan and Lombardy. — “We presented the word categories whose frequencies significantly changed after the lockdown both in Wuhan and Lombardy in Table 3 , including home and discrepancy . In both Wuhan and Lombardy, the frequencies of home and discrepancy words increased after a lockdown.” | content_tags.findings[0] |
| finding | Wuhan showed significant changes in 39 of 51 shared LIWC categories. — “In this study, we compared the word frequencies of 51 LIWC categories before lockdown with after lockdown in Wuhan. Results showed that the frequencies of 39 word categories were statistically significantly different before and after Wuhan lockdown. We identified 16 out of 39 significant categories with absolute values of effect size greater than 0.2, including function words (e.g., I, we ), relative words ( motion, time ), personal concerns words ( home, money, religion ), affective process words ( negative emotion, affect ), social words ( humans, social ), and cognitive mechanism words (e.g., certain, inhibition). As shown in Table 1 , the first-person plural pronoun is of high effect size ( p < 0.001, effect size d = 0.674), which means users used more words of the first-person plural pronoun significantly after the lockdown. In addition, Weibo users mentioned more in religion , social , negative emotion , home , affect after Wuhan lockdown. Meanwhile, we found significant decreases in the frequencies of some word categories, such as motion, I, money, and time after the lockdown.” | content_tags.findings[1] |
| finding | Lombardy showed significant changes in ten shared categories, including reduced anxiety-word use. — “We compared the word frequencies of the 51 LIWC categories before and after Lombardy lockdown (8 March 2020). Results showed that the frequencies of 10-word categories were significantly changed. Among them, the number of word categories with absolute values of effect size greater than 0.2 is five-word categories, including personal concerns words ( leisure, home ), affective process words ( anxiety ), and cognitive mechanism words ( discrepancy, possibility ). As shown in Table 2 , there are increases in the frequencies of discrepancy, home, leisure , and possibility . Meanwhile, we observed significant decrease in the frequency of anxiety after the lockdown.” | content_tags.findings[2] |
| finding | Pronoun changes in Wuhan were interpreted as a shift of attention from self toward other people and groups. — “In Wuhan, the uses of the first-person plural pronouns, second-person plural pronoun increase after a lockdown, while the use of first-person singular pronoun decreases. Previous reports confirm that first-person singular pronouns show attention to the self, whereas most other pronouns suggest attention to other individuals [ 36 ]. Moreover, “we” implies a sense of group identity sometimes [ 37 ]. Results suggest that people switch their attention from themselves to others and the communities after the lockdown. Besides, the increased use of “we” indicates that people focus more on the group, become more united, and share more group identity after a lockdown, which is consistent with some researchers’ opinions [ 1 ]. China has a collectivistic culture, and Italy has an individualistic culture [ 38 ]. Results show that the increased use of other pronouns and decreased use of first-person singular pronouns suggest a collectivistic culture in China. At the same time, the absence of such a consequence in Lombardy might be related to the individualist culture. Researchers find that people sharing collectivist values stress more communal coping as a resource to cope with collective traumatic events [ 39 ], which is consistent with our research conducted in the context of the lockdown.” | content_tags.findings[3] |
| finding | Emotion-word use increased in Wuhan and was interpreted as deeper immersion in negative emotion. — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | content_tags.findings[4] |
| finding | Tentative-language use increased in Lombardy but not Wuhan. — “We find significant changes in three-word categories in Lombardy, including tentative , anxiety , and leisure words. The use of tentative words increases after a lockdown in Lombardy. The previous study shows that people may use tentative language (e.g., maybe, perhaps, guess) when they feel uncertain or insecure about their topic [ 22 ]. Our findings suggest that people tend to use tentative words during the lockdown. Losing direct social contacts during the lockdown contributes to make residents feel losses of recreation, freedoms, and supports [ 1 ]. Such a sense of loss means losing control of their healthy life, and people are likely to feel uncertain about the upcoming situation. Tweets reveal that people in Lombardy express such feelings on social media. However, we do not observe such change in Wuhan, suggesting that people in Wuhan do not convey the emotions of uncertainty in their posts on Weibo.” | content_tags.findings[5] |
| finding | Wuhan social-word use increased, consistent with greater focus on social concerns and support. — “The increased use of social words in Wuhan after the lockdown suggests the focus on social concerns and social support [ 22 ]. Social support can make people feel better about their situation and reinforce the belief that they have access to support resources [ 16 , 42 ]. Thus, seeking social support is considered adaptive for people during a lockdown. In contrast, we do not observe such a change in Lombardy.” | content_tags.findings[6] |
| finding | The number and type of changed linguistic categories differed substantially between the regions. — “This study examined the changes in psycholinguistic features before and after a lockdown in Wuhan and Lombardy. We compared the differences in frequencies of LIWC word categories before and after lockdown and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdown in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdown. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdown on individuals in the two countries, and contribute to the cultural-based psychological responses.” | content_tags.findings[7] |
| quantity | 1.16 million — active users in the source Weibo data pool — “The Chinese samples are from the Weibo data pool containing 1.16 million active Weibo users [ 28 ]. In this study, we selected active Weibo users from the data pool by the following criteria:” | content_tags.quantities[0] |
| quantity | 850 — selected Wuhan Weibo users in the final feature dataset — “We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdown” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdown” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWC dictionary containing 51 common word categories to extract psycholinguistic features and calculated word frequencies of each word category in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.” | content_tags.quantities[1] |
| quantity | 3,650,380 — initial tweets acquired for Lombardy sampling — “We sampled Italian Twitter users’ messages as our Twitter data. We downloaded Tweets of users whose location authentication is Lombardia, Italy. There are 3,650,380 tweets acquired. We then selected Italian Twitter users as follows:” | content_tags.quantities[2] |
| quantity | 120 — Lombardy Twitter users retained after filtering — “We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdown” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdown” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdown” or “after lockdown” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word category.” | content_tags.quantities[3] |
| quantity | 51 — harmonized LIWC word categories analyzed across both languages — “Some word categories are unique in SCLIWC or Italian LIWC. By comparing, there are 26 word categories only existing in SCLIWC, such as quantity unit, interjunction , and tense mark words . In Italian, people conjugate verbs when they follow different subjects. Moreover, people can infer the subject of the sentence from verb conjugation. As the subject in a sentence is dropped sometimes, conjugations ( I_verb, We_verb, You_verb, You_plural_verb, HeShe_verb, They_verb ) can reveal the use of pronouns more accurately compared to pronouns ( I, We, You, You plural, HeShe, Them ). Thus, we regard the use of conjugations, the same as pronouns in our study. Additionally, we found that 28 word categories only exist in Italian LIWC. In this study, we kept the common word categories in both SCLIWC and Italian LIWC and got 51 common word categories for further analysis.” | content_tags.quantities[4] |
| quantity | 39 — categories with significant pre-post frequency differences in Wuhan — “In this study, we compared the word frequencies of 51 LIWC categories before lockdown with after lockdown in Wuhan. Results showed that the frequencies of 39 word categories were statistically significantly different before and after Wuhan lockdown. We identified 16 out of 39 significant categories with absolute values of effect size greater than 0.2, including function words (e.g., I, we ), relative words ( motion, time ), personal concerns words ( home, money, religion ), affective process words ( negative emotion, affect ), social words ( humans, social ), and cognitive mechanism words (e.g., certain, inhibition). As shown in Table 1 , the first-person plural pronoun is of high effect size ( p < 0.001, effect size d = 0.674), which means users used more words of the first-person plural pronoun significantly after the lockdown. In addition, Weibo users mentioned more in religion , social , negative emotion , home , affect after Wuhan lockdown. Meanwhile, we found significant decreases in the frequencies of some word categories, such as motion, I, money, and time after the lockdown.” | content_tags.quantities[5] |
| variable | lockdown period (within-user exposure contrasting posts before and after each regional lockdown) — “We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdown” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdown” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWC dictionary containing 51 common word categories to extract psycholinguistic features and calculated word frequencies of each word category in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.” | content_tags.variables[0] |
| variable | use of emotion words (linguistic indicator of affective immersion) — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | content_tags.variables[4] |
| definition | word category — “Existing studies have widely used the Language Inquiry and Word Count (LIWC) and confirmed it as a valid tool for psychometric analysis [ 22 ]. The LIWC dictionary has multiple versions of different languages, including English [ 23 ], French [ 24 ], Italian [ 25 ], and Dutch [ 26 ]. The LIWC dictionary includes many word categories of linguistic features that are related to mental processes and human behaviors [ 22 ]. For example, the word category of personal pronouns reflects attentional allocation [ 22 ].” | content_tags.definitions[0] |
| definition | attentional allocation — “Existing studies have widely used the Language Inquiry and Word Count (LIWC) and confirmed it as a valid tool for psychometric analysis [ 22 ]. The LIWC dictionary has multiple versions of different languages, including English [ 23 ], French [ 24 ], Italian [ 25 ], and Dutch [ 26 ]. The LIWC dictionary includes many word categories of linguistic features that are related to mental processes and human behaviors [ 22 ]. For example, the word category of personal pronouns reflects attentional allocation [ 22 ].” | content_tags.definitions[1] |
| definition | tentative language — “We find significant changes in three-word categories in Lombardy, including tentative , anxiety , and leisure words. The use of tentative words increases after a lockdown in Lombardy. The previous study shows that people may use tentative language (e.g., maybe, perhaps, guess) when they feel uncertain or insecure about their topic [ 22 ]. Our findings suggest that people tend to use tentative words during the lockdown. Losing direct social contacts during the lockdown contributes to make residents feel losses of recreation, freedoms, and supports [ 1 ]. Such a sense of loss means losing control of their healthy life, and people are likely to feel uncertain about the upcoming situation. Tweets reveal that people in Lombardy express such feelings on social media. However, we do not observe such change in Wuhan, suggesting that people in Wuhan do not convey the emotions of uncertainty in their posts on Weibo.” | content_tags.definitions[2] |
| definition | anxiety words — “The use of anxiety words decreases in Lombardy. Anxiety reveals self-reported stress [ 34 ]. Our results imply that people feel less stressed after Lombardy lockdown. However, people do not experience any change of stress in Wuhan. Researchers find that unrealistic optimism is more evident for European North Americans [ 13 ], which might be related to the different responses in the level of stress between Lombardy and Wuhan after the lockdown. However, our results are not consistent with existing studies [ 12 , 35 ]. Rossi and colleagues consider that the strict measures of the lockdown in Italy serve as an unprecedented stressful event [ 12 ]. Besides, Ahmed and colleagues find that 29% of respondents report different levels of anxiety related to lockdown at home in China [ 35 ]. Such differences could be due to different research methods, design, measurements, and timeframe used in the study.” | content_tags.definitions[3] |
| definition | immersion — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | content_tags.definitions[4] |
| definition | content word categories — “Table 1 shows increases in the uses of religion and humans words, while decreases in the uses of money and time words after Wuhan lockdown. Content word categories explicitly reveal where individuals are focusing, including death, sex, and money [ 22 ]. Moreover, our results suggest people focus more on humans and religion, while less on money and time during the lockdown. The previous study finds that religion can bring more positive and comforting emotions, and people tend to use it when suffering from emergencies such as stress or death [ 43 ]. The increases in the use of religion words suggest an adaptive behavior during the lockdown. Moreover, the decreased use of money words may relate to fewer transactions under strict restrictions. In contrast, we do not identify any changes in these word categories among Lombardy Tweets. This result in Lombardy Tweets suggests that residents in Lombardy do not change their focus level on religion, human, time, and money after lockdown.” | content_tags.definitions[5] |
| concept | LIWC — not defined in this paper's own words; multilingual dictionary for extracting psycholinguistic word categories — “Existing studies have widely used the Language Inquiry and Word Count (LIWC) and confirmed it as a valid tool for psychometric analysis [ 22 ]. The LIWC dictionary has multiple versions of different languages, including English [ 23 ], French [ 24 ], Italian [ 25 ], and Dutch [ 26 ]. The LIWC dictionary includes many word categories of linguistic features that are related to mental processes and human behaviors [ 22 ]. For example, the word category of personal pronouns reflects attentional allocation [ 22 ].” | concepts[0] |
| concept | psycholinguistic feature — not defined in this paper's own words; word-category frequency used as a psychological-state indicator — “The present study uses the Chinese version of LIWC and Italian version of LIWC to extract the psycholinguistic features from social media users’ posts. Examinations of the features allow us to access the changes of psychological status before and after the lockdown in Wuhan and Lombardy.” | concepts[1] |
| concept | attentional allocation — not defined in this paper's own words; distribution of focus inferred from pronoun use — “Existing studies have widely used the Language Inquiry and Word Count (LIWC) and confirmed it as a valid tool for psychometric analysis [ 22 ]. The LIWC dictionary has multiple versions of different languages, including English [ 23 ], French [ 24 ], Italian [ 25 ], and Dutch [ 26 ]. The LIWC dictionary includes many word categories of linguistic features that are related to mental processes and human behaviors [ 22 ]. For example, the word category of personal pronouns reflects attentional allocation [ 22 ].” | concepts[2] |
| concept | lockdown — not defined in this paper's own words; regional mobility restriction used as the before-after event — “The “first case” of COVID-19 was identified in Wuhan [ 10 ], which was the epicenter of the coronavirus outbreak in China. To stop the spread of COVID-19, China declared the lockdown of Wuhan on 23 January 2020, which was the first city placed on lockdown during this pandemic in China and affected over 11 million people [ 11 ]. In Europe, Italy was the first country facing the pandemic [ 12 ] and taking actions (e.g., banned flights) [ 13 ]. Lombardy in Italy was the most affected area by COVID-19 [ 14 ]. On 8 March 2020, the Italian government announced a quarantine zone that covered most of northern Italy, including Lombardy. Lombardy had a population of over 10 million, which was comparable with Wuhan. This quarantine measure was considered as the most aggressive response taken in any region beyond China [ 15 ]. Taken all these together, we chose Wuhan and Lombardy as research regions to investigate the impacts of the lockdown.” | concepts[3] |
| concept | discrepancy words — not defined in this paper's own words; cognitive-process category interpreted as sense-making effort — “The frequency of discrepancy words increases after the lockdown in Wuhan and Lombardy. Besides, we observe the increased use of inhibition and certain words after the lockdown in Wuhan. Previous study suggests that the uses of discrepancy, Inhibition, and certain words reflect the change of degree of cognitive processing [ 32 ]. Furthermore, cognitive processing indicates that individuals make efforts to make sense of the environment [ 32 ]. Residents in Wuhan and Lombardy attempt to figure out what has happened after the lockdown. Thus, they could adjust their attitudes and lifestyles to accommodate new circumstances during the COVID-19 pandemic.” | concepts[4] |
| concept | tentative words — not defined in this paper's own words; uncertainty-marking language — “We find significant changes in three-word categories in Lombardy, including tentative , anxiety , and leisure words. The use of tentative words increases after a lockdown in Lombardy. The previous study shows that people may use tentative language (e.g., maybe, perhaps, guess) when they feel uncertain or insecure about their topic [ 22 ]. Our findings suggest that people tend to use tentative words during the lockdown. Losing direct social contacts during the lockdown contributes to make residents feel losses of recreation, freedoms, and supports [ 1 ]. Such a sense of loss means losing control of their healthy life, and people are likely to feel uncertain about the upcoming situation. Tweets reveal that people in Lombardy express such feelings on social media. However, we do not observe such change in Wuhan, suggesting that people in Wuhan do not convey the emotions of uncertainty in their posts on Weibo.” | concepts[5] |
| concept | emotion words — not defined in this paper's own words; affective-process category used to index immersion — “Holmes and colleagues find that higher levels of the use of emotion words indicate more immersion in the negative event [ 40 ]. In the study, we find that a higher degree of immersion [ 22 ] evidenced by the frequent use of emotion words ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersion in negative emotions after the lockdown. However, we do not observe such a situation in Lombardy.” | concepts[6] |
| concept | bias existing in two different platforms — not defined in this paper's own words; systematic Weibo-Twitter differences in hashtag and positive-content practices — “There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platforms possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.” | concepts[7] |
| concept | group identity — not defined in this paper's own words; collective orientation inferred from plural-pronoun use — “In Wuhan, the uses of the first-person plural pronouns, second-person plural pronoun increase after a lockdown, while the use of first-person singular pronoun decreases. Previous reports confirm that first-person singular pronouns show attention to the self, whereas most other pronouns suggest attention to other individuals [ 36 ]. Moreover, “we” implies a sense of group identity sometimes [ 37 ]. Results suggest that people switch their attention from themselves to others and the communities after the lockdown. Besides, the increased use of “we” indicates that people focus more on the group, become more united, and share more group identity after a lockdown, which is consistent with some researchers’ opinions [ 1 ]. China has a collectivistic culture, and Italy has an individualistic culture [ 38 ]. Results show that the increased use of other pronouns and decreased use of first-person singular pronouns suggest a collectivistic culture in China. At the same time, the absence of such a consequence in Lombardy might be related to the individualist culture. Researchers find that people sharing collectivist values stress more communal coping as a resource to cope with collective traumatic events [ 39 ], which is consistent with our research conducted in the context of the lockdown.” | concepts[8] |
Full text — tagged
Examining the Impact of COVID-19 Lockdownterm·dom in Wuhan and Lombardy: A Psycholinguistic Analysis on Weibo and Twitter Abstract Many countries are taking strict quarantine policies to prevent the rapid spread of COVID-19 (Corona Virus Disease 2019) around the world, such as city lockdownterm·dom. Cities in China and Italy were locked down in the early stage of the pandemic. The present study aims to examine and compare the impact of COVID-19 lockdownterm·dom on individuals’ psychological states in China and Italy. We achieved the aim by (1) sampling Weibo users (geo-location = Wuhan, China) and Twitter users (geo-location = Lombardy, Italy); (2) fetching all the users’ published posts two weeks before and after the lockdownterm·dom in each region (e.g., the lockdownterm·dom date of Wuhan was 23 January 2020); (3) extracting the psycholinguistic featuresterm·dom of these posts using the Simplified Chinese and Italian version of Language Inquiry and Word Count (LIWCterm·mth) dictionary; and (4) conducting Wilcoxon tests to examine the changes in the psycholinguistic characteristics of the posts before and after the lockdownterm·dom in Wuhan and Lombardy, respectively. Results showed that individuals focused more on “home”, and expressed a higher level of cognitive process after a lockdownterm·dom in both Wuhan and Lombardy. Meanwhile, the level of stress decreased, and the attention to leisure increased in Lombardy after the lockdownterm·dom. The attention to group, religion, and emotions became more prevalent in Wuhan after the lockdownterm·dom. Findings provide decision-makers timely evidence on public reactions and the impacts on psychological states in the COVID-19 context, and have implications for evidence-based mental health interventions in two countries.
Keywords: impact of COVID-19 lockdownterm·dom, public health emergencies, psycholinguistic analysis, psychological states 1. Introduction The COVID-19 (Corona Virus Disease 2019) pandemic is a global health emergency that is having a profound impact on the physical and mental health of people [ 1 , 2 , 3 ]. Many countries have taken strict quarantine measures as an intervention: cities locked down, school closure, mass gathering ban, public event prohibition, and self-isolation. A study conducted in China shows that lockdownterm·dom has been effective in postponing the spread of COVID-19 [ 4 ]. However, strict quarantine interventions may have negative impacts on mental health [ 5 , 6 ]. It is essential to investigate the psychological effects of the lockdownterm·dom which could make an influence on the execution of measures on epidemic containment. Studies find that public reactions to SARS (Severe Acute Respiratory Syndrome) in 2003 and the Ebola virus disease in 2014 have impeded infection control to an extent [ 7 , 8 ]. Moreover, quarantine measures are making psychosocial impact on individuals more severe [ 9 ]. This study intends to explore how the lockdownterm·dom affects the psychological states.
The “first case” of COVID-19 was identified in Wuhan [ 10 ], which was the epicenter of the coronavirus outbreak in China. To stop the spread of COVID-19, China declared the lockdownterm·dom of Wuhan on 23 January 2020, which was the first city placed on lockdownterm·dom during this pandemic in China and affected over 11 million people [ 11 ]. In Europe, Italy was the first country facing the pandemic [ 12 ] and taking actions (e.g., banned flights) [ 13 ]. Lombardy in Italy was the most affected area by COVID-19 [ 14 ]. On 8 March 2020, the Italian government announced a quarantine zone that covered most of northern Italy, including Lombardy. Lombardy had a population of over 10 million, which was comparable with Wuhan. This quarantine measure was considered as the most aggressive response taken in any region beyond China [ 15 ]. Taken all these together, we chose Wuhan and Lombardy as research regions to investigate the impacts of the lockdownterm·dom.
Recent studies used the self-report questionnaire approach to examine the psychological responses during the lockdownterm·dom in different countries, including Italy, India, and China [ 12 , 16 , 17 ]. However, these studies rely on retrospective and time-lagged surveys and interviews. These approaches have limitations in accessing psychological statuses before the lockdownterm·dom. That is, there is recall bias inevitably when people are required to recall a past period in the retrospective study.
Social media plays a vital role in recording the reactions, opinions, and mental health features of social media users [ 18 ] Previous studies suggest that the language use and psychosocial expressions on social media data provide indicators of mental health [ 19 , 20 , 21 ]. In China, Sina Weibo is the leading social media service provider. Upon the end of 2019, the number of daily active users of Sina Weibo reached 222 million. Twitter is one of the most-used social media platforms in Italy. Weibo and Twitter provide vast amounts of user’s online behavioral records for researchers. Although there are some differences between Weibo and Twitter when comparing the functions and other features of platforms, they both serve as the online environment of expression and communication, providing us features of contents related to this study. Thus, we collected Chinese social media data from Weibo and used Twitter to acquire Italian social media data.
Existing studies have widely used the Language Inquiry and Word Count (LIWCterm·mth) and confirmed it as a valid tool for psychometric analysis [ 22 ]. The LIWCterm·mth dictionary has multiple versions of different languages, including English [ 23 ], French [ 24 ], Italian [ 25 ], and Dutch [ 26 ]. The LIWCterm·mth dictionary includes many word categories of linguistic features that are related to mental processes and human behaviors [ 22 ]. For example, the word categoryterm·def of personal pronouns reflects attentional allocationterm·dom [ 22 ].
In this study, we used the simplified Chinese version of LIWCterm·mth and Italian LIWCterm·mth to measure people’s psychological status before and after the lockdownterm·dom in Wuhan and Lombardy, respectively. By using psycholinguistic analysis, we aim to identify the psychological effects of the lockdownterm·dom on individuals in Wuhan and Lombardy.
2. Materials and Methods We downloaded active user’s posts from Weibo and Twitter as our dataset. The research protocol was approved in advance by the Ethics Committee of the Institute of Psychology, Chinese Academy of Sciences (approved number: H15009 ).
We extracted the linguistic features using the Simplified Chinese LIWCterm·mth dictionary (SCLIWC) [ 27 ] and the Italian LIWCterm·mth dictionary [ 25 ]. Given both SCLIWC and Italian LIWCterm·mth share LIWCterm·mth dictionary structure, there are many common words in SCLIWC and LIWCterm·mth. To make the result of Wuhan and Lombardy comparable, we only analyzed the common word categories between SCLIWC and Italian LIWCterm·mth. The selection procedure of common categories is as follows: A native Italian speaker who is fluent in English translated the names of Italian LIWCterm·mth word categories into English.
We translated the Chinese names of SCLIWC word categories into English.
We selected the common names between two translation versions. As for the names sharing similar meanings, such as “tentative” from SCLIWC and “possibility” from Italian LIWCterm·mth, we checked the meaning of words belonging to this word categoryterm·def in Italian LIWCterm·mth and SCLIWC to determine whether the two names represented the same kind of word categoryterm·def.
2.1. Dictionary Processing Some word categories are unique in SCLIWC or Italian LIWCterm·mth. By comparing, there are 26 word categories only existing in SCLIWC, such as quantity unit, interjunction , and tense mark words . In Italian, people conjugate verbs when they follow different subjects. Moreover, people can infer the subject of the sentence from verb conjugation. As the subject in a sentence is dropped sometimes, conjugations ( I_verb, We_verb, You_verb, You_plural_verb, HeShe_verb, They_verb ) can reveal the use of pronouns more accurately compared to pronouns ( I, We, You, You plural, HeShe, Them ). Thus, we regard the use of conjugations, the same as pronouns in our study. Additionally, we found that 28 word categories only exist in Italian LIWCterm·mth. In this study, we kept the common word categories in both SCLIWC and Italian LIWCterm·mth and got 51 common word categories for further analysis.
2.2. Weibo Data The Chinese samples are from the Weibo data pool containing 1.16 million active Weibo users [ 28 ]. In this study, we selected active Weibo users from the data pool by the following criteria: Published at least one original post on average each day from 9 January 2020 to 5 February 2020 (i.e., two weeks before and after the lockdownterm·dom); Individual users only, excluding any organizations; Locate at “Wuhan, Hubei” by the geo-location in the user profile.
We finally got 850 Weibo users and downloaded their posts published from 9 January 2020 to 5 February 2020. For each Weibo user, we divided the posts into two groups. For example, all posts published before the date 23 January 2020, are labeled as “before lockdownterm·dom” group. In contrast, those posts published after the data 23 January 2020 (23 January included) are labeled as “after lockdownterm·dom” group. We employed the TextMind system to extract linguistic features [ 29 ] in each of the two groups for all sampled Weibo users. Then, we used the LIWCterm·mth dictionary containing 51 common word categories to extract psycholinguistic featuresterm·dom and calculated word frequencies of each word categoryterm·def in the dictionary. The final dataset included the word frequencies of two groups from 850 Weibo users.
2.3. Twitter Data We sampled Italian Twitter users’ messages as our Twitter data. We downloaded Tweets of users whose location authentication is Lombardia, Italy. There are 3,650,380 tweets acquired. We then selected Italian Twitter users as follows: Published at least one original tweet (not retweet) from 23 February 2020 to 21 March 2020 (that is, two weeks before and after the lockdownterm·dom); All tweets in Italian only.
We acquired 14,269 Tweets from 188 unique Twitter users. We divided these Tweets into two groups as well. We gathered each user’s tweets and labeled the Tweets posted before 8 March 2020, as “before lockdownterm·dom” Tweets and Tweets posted after 8 March 2020 (March 8 included) as “after lockdownterm·dom” Tweets, respectively. We filtered out the users if only emoji, numbers, web links, “@” and “#” were published in either “before lockdownterm·dom” or “after lockdownterm·dom” tweets. We finally acquired 120 Twitter users. Then, we extracted every user’s linguistic features from “before” and “after” tweets by using the same dictionary used in Weibo data and calculated word frequencies of each word categoryterm·def.
We conducted Wilcoxon tests to examine the differences between linguistic characteristics before and after the lockdownterm·dom. SPSS (Statistical Product and Service Solutions) 26.0 (International Business Machines Corporation, Armonk, NY, USA). Released 2019. IBM SPSS Statistics for Macintosh, Version 26.0. was used during data analysis, which was published by IBM (International Business Machines Corporation, Armonk, NY, USA).
3. Results 3.1. Wuhan Weibo Users In this study, we compared the word frequencies of 51 LIWCterm·mth categories before lockdownterm·dom with after lockdownterm·dom in Wuhan. Results showed that the frequencies of 39 word categories were statistically significantly different before and after Wuhan lockdownterm·dom. We identified 16 out of 39 significant categories with absolute values of effect size greater than 0.2, including function words (e.g., I, we ), relative words ( motion, time ), personal concerns words ( home, money, religion ), affective process words ( negative emotion, affect ), social words ( humans, social ), and cognitive mechanism words (e.g., certain, inhibition). As shown in Table 1 , the first-person plural pronoun is of high effect size ( p < 0.001, effect size d = 0.674), which means users used more words of the first-person plural pronoun significantly after the lockdownterm·dom. In addition, Weibo users mentioned more in religion , social , negative emotion , home , affect after Wuhan lockdownterm·dom. Meanwhile, we found significant decreases in the frequencies of some word categories, such as motion, I, money, and time after the lockdownterm·dom.
Word categories with significant differences between “before” and “after” in Weibo ( n = 850).
3.2. Lombardy Twitter Users We compared the word frequencies of the 51 LIWCterm·mth categories before and after Lombardy lockdownterm·dom (8 March 2020). Results showed that the frequencies of 10-word categories were significantly changed. Among them, the number of word categories with absolute values of effect size greater than 0.2 is five-word categories, including personal concerns words ( leisure, home ), affective process words ( anxiety ), and cognitive mechanism words ( discrepancy, possibility ). As shown in Table 2 , there are increases in the frequencies of discrepancy, home, leisure , and possibility . Meanwhile, we observed significant decrease in the frequency of anxiety after the lockdownterm·dom.
Words with significant changes between “before lockdownterm·dom” and “after lockdownterm·dom” in Lombardy ( n = 120).
3.3. Comparison between Wuhan and Lombardy We presented the word categories whose frequencies significantly changed after the lockdownterm·dom both in Wuhan and Lombardy in Table 3 , including home and discrepancy . In both Wuhan and Lombardy, the frequencies of home and discrepancy wordsterm·dom increased after a lockdownterm·dom.
Words with significant changes in both Wuhan and Lombardy.
4. Discussion The present study uses the Chinese version of LIWCterm·mth and Italian version of LIWCterm·mth to extract the psycholinguistic featuresterm·dom from social media users’ posts. Examinations of the features allow us to access the changes of psychological status before and after the lockdownterm·dom in Wuhan and Lombardy.
4.1. Similarities between Wuhan and Lombardy The frequencies of some word categories increase in both Wuhan and Lombardy after the lockdownterm·dom, including discrepancy and home words. These linguistic features imply that social media users’ psychological states were impacted after the COVID-19 lockdownterm·dom, in both Wuhan and Lombardy.
The increased use of home words is related to mobility control after the lockdownterm·dom in Wuhan and Lombardy. Researchers estimated that mobility and social contacts in China during the lockdownterm·dom dropped about 80%, concerning a baseline set on 1 January 2020 [ 30 ]. Moreover, Google reported a 23% increase in residential location activity in Lombardy during lockdownterm·dom compared to baseline [ 31 ]. These reports indicate people spend more time at home and spend less time outdoors during a lockdownterm·dom, which is consistent with more use of home words.
The frequency of discrepancy wordsterm·dom increases after the lockdownterm·dom in Wuhan and Lombardy. Besides, we observe the increased use of inhibition and certain words after the lockdownterm·dom in Wuhan. Previous study suggests that the uses of discrepancy, Inhibition, and certain words reflect the change of degree of cognitive processing [ 32 ]. Furthermore, cognitive processing indicates that individuals make efforts to make sense of the environment [ 32 ]. Residents in Wuhan and Lombardy attempt to figure out what has happened after the lockdownterm·dom. Thus, they could adjust their attitudes and lifestyles to accommodate new circumstances during the COVID-19 pandemic.
4.2. Differences between Wuhan and Lombardy We observe there are some differences between Wuhan and Lombardy after a lockdownterm·dom in the use of LIWCterm·mth word categories.
We find significant changes in three-word categories in Lombardy, including tentative , anxiety , and leisure words. The use of tentative wordsterm·dom increases after a lockdownterm·dom in Lombardy. The previous study shows that people may use tentative languageterm·def (e.g., maybe, perhaps, guess) when they feel uncertain or insecure about their topic [ 22 ]. Our findings suggest that people tend to use tentative wordsterm·dom during the lockdownterm·dom. Losing direct social contacts during the lockdownterm·dom contributes to make residents feel losses of recreation, freedoms, and supports [ 1 ]. Such a sense of loss means losing control of their healthy life, and people are likely to feel uncertain about the upcoming situation. Tweets reveal that people in Lombardy express such feelings on social media. However, we do not observe such change in Wuhan, suggesting that people in Wuhan do not convey the emotions of uncertainty in their posts on Weibo.
Our results show that Twitter users in Lombardy use more leisure words in their posts after the lockdownterm·dom. The increased use of leisure words implies more focus on leisure activities after a lockdownterm·dom in Lombardy. According to the news reports from CNBC (Consumer News and Business Channel), Italians turn to music to boost morale during lockdownterm·dom [ 33 ], which might be expressed in the use of leisure in Tweets. On the contrary, we do not find the same change in the use of leisure words in Wuhan. With the rapid growth of the pandemic, some people might focus more attention on the latest news of this disease on Weibo and discuss less about leisure after the lockdownterm·dom. Moreover, some people may talk more about leisure and recreation after the lockdownterm·dom, considering that the Lunar New Year holiday was in the lockdown periodterm·var (25 January 2020, is the Spring Festival in China). Considering these two facets, we may find it reasonable to observe no change in the use of leisure words in Wuhan.
The use of anxiety wordsterm·def decreases in Lombardy. Anxiety reveals self-reported stress [ 34 ]. Our results imply that people feel less stressed after Lombardy lockdownterm·dom. However, people do not experience any change of stress in Wuhan. Researchers find that unrealistic optimism is more evident for European North Americans [ 13 ], which might be related to the different responses in the level of stress between Lombardy and Wuhan after the lockdownterm·dom. However, our results are not consistent with existing studies [ 12 , 35 ]. Rossi and colleagues consider that the strict measures of the lockdownterm·dom in Italy serve as an unprecedented stressful event [ 12 ]. Besides, Ahmed and colleagues find that 29% of respondents report different levels of anxiety related to lockdownterm·dom at home in China [ 35 ]. Such differences could be due to different research methods, design, measurements, and timeframe used in the study.
Some word categories are changing significantly after the lockdownterm·dom only in Wuhan. The uses of first-person plural pronouns, second-person plural pronouns, religion, social, negative emotions, humans, certainty, affect, inhibition, and prepositions words increase. In contrast, the uses of motion, first-person singular pronouns, time, and money words decrease after the lockdownterm·dom in Wuhan.
In Wuhan, the uses of the first-person plural pronouns, second-person plural pronoun increase after a lockdownterm·dom, while the use of first-person singular pronoun decreases. Previous reports confirm that first-person singular pronouns show attention to the self, whereas most other pronouns suggest attention to other individuals [ 36 ]. Moreover, “we” implies a sense of group identityterm·dom sometimes [ 37 ]. Results suggest that people switch their attention from themselves to others and the communities after the lockdownterm·dom. Besides, the increased use of “we” indicates that people focus more on the group, become more united, and share more group identityterm·dom after a lockdownterm·dom, which is consistent with some researchers’ opinions [ 1 ]. China has a collectivistic culture, and Italy has an individualistic culture [ 38 ]. Results show that the increased use of other pronouns and decreased use of first-person singular pronouns suggest a collectivistic culture in China. At the same time, the absence of such a consequence in Lombardy might be related to the individualist culture. Researchers find that people sharing collectivist values stress more communal coping as a resource to cope with collective traumatic events [ 39 ], which is consistent with our research conducted in the context of the lockdownterm·dom.
Holmes and colleagues find that higher levels of the use of emotion wordsterm·var indicate more immersionterm·def in the negative event [ 40 ]. In the study, we find that a higher degree of immersionterm·def [ 22 ] evidenced by the frequent use of emotion wordsterm·var ( negative emotion and affective process words). Therefore, people in Wuhan might get more emotional and are at a deeper level of immersionterm·def in negative emotions after the lockdownterm·dom. However, we do not observe such a situation in Lombardy.
Besides, we also find a decrease in the use of motion words after a lockdownterm·dom in Wuhan. Our results are consistent with the previous mobility study of Wuhan [ 41 ], suggesting that stringent mobility control leads to the reduction of movement in Wuhan. Google’s location mobility report in Lombardy shows an 85% decrease in activities at transit stations, a 57% drop in activities at workplaces, an 86% drop in activities at parks, and a 94% drop in activities at retail and recreation from 15 March 2020 to 26 April 2020. However, our results do not identify a significant change of mobility in the use of motion words in Lombardy.
The increased use of social words in Wuhan after the lockdownterm·dom suggests the focus on social concerns and social support [ 22 ]. Social support can make people feel better about their situation and reinforce the belief that they have access to support resources [ 16 , 42 ]. Thus, seeking social support is considered adaptive for people during a lockdownterm·dom. In contrast, we do not observe such a change in Lombardy.
Table 1 shows increases in the uses of religion and humans words, while decreases in the uses of money and time words after Wuhan lockdownterm·dom. Content word categoriesterm·def explicitly reveal where individuals are focusing, including death, sex, and money [ 22 ]. Moreover, our results suggest people focus more on humans and religion, while less on money and time during the lockdownterm·dom. The previous study finds that religion can bring more positive and comforting emotions, and people tend to use it when suffering from emergencies such as stress or death [ 43 ]. The increases in the use of religion words suggest an adaptive behavior during the lockdownterm·dom. Moreover, the decreased use of money words may relate to fewer transactions under strict restrictions. In contrast, we do not identify any changes in these word categories among Lombardy Tweets. This result in Lombardy Tweets suggests that residents in Lombardy do not change their focus level on religion, human, time, and money after lockdownterm·dom.
Besides, we find an increase in the use of prepositions . Previous research shows that prepositions signal more complex expression and detailed information about a topic [ 44 ]. The increased use of prepositions in Wuhan indicates broader and more in-depth discussions that occurred on Weibo after lockdownterm·dom. However, such a change is not identified in Lombardy.
Study findings have implications for decision-makers, public health authorities, and practitioners. First, considering the efforts of adjusting to the changing environment in both Wuhan and Lombardy after the lockdownterm·dom, decision-makers should ensure the supply chain functions as usual to ensure people’s confidence in having the control of their lives. Besides, public health authorities and practitioners could adjust their focus of service given the changes in residents’ attention after lockdownterm·dom. For example, people in Wuhan expressed more stress and negative emotions, public health authorities and practitioners should take interventions to comfort them and relieve stress, such as the online consulting service and indoor activities. Notably, the support for individuals with pre-existing mental or physical health issues is also needed. Meanwhile, people did not show significant stress in Lombardy. Public health communities and practitioners might focus more on the popularization of pandemic prevention knowledge and the reinforcement of protection awareness.
There are several limitations. First, our samples were from selected active social media users only. The results have a limitation in generalizing to the whole population. Second, language differences exist between Chinese and Italian. While processing Italian text, some inevitable errors may occur because of the apostrophe. Third, we do not have access to the users’ IP, and location authentication is self-reported. There are some studies also applying self-reported location authentication to identify users’ locations [ 45 ]. Fourth, the bias existing in two different platformsterm·dom possibly influences the results of our study. Twitter users generally use more hashtags than Weibo users, which shows that Twitter users seem to be more eager to publicize their posts [ 46 ]. In addition, Weibo users have a stronger tendency to post positive content compared to Twitter users [ 46 ]. Considering these differences between Twitter and Weibo, future studies should find methods to deal with these differences to avoid biases when employing data from Weibo and Twitter.
5. Conclusions This study examined the changes in psycholinguistic featuresterm·dom before and after a lockdownterm·dom in Wuhan and Lombardy. We compared the differences in frequencies of LIWCterm·mth word categories before and after lockdownterm·dom and found that the number of word categories whose frequencies were significantly changed is more in Wuhan than in Lombardy. We found significant changes in the use of function words, relative words, personal concerns words, affective process words, social words, and cognitive mechanism words among Wuhan users’ posts. We also found significant changes in the frequencies of personal concerns words, affective process words, and cognitive mechanism words in Lombardy. Individuals focus more on home and express more levels of the cognitive process after a lockdownterm·dom in both Wuhan and Lombardy. In Lombardy, the level of stress decreases, the use of leisure increases. In Wuhan, people convey more emotion expressions, more feelings of uncertainty, and more focus on groups after the lockdownterm·dom. Results inform decision-makers, public health authorities, and practitioners the potentially different impacts of city lockdownterm·dom on individuals in the two countries, and contribute to the cultural-based psychological responses.
Acknowledgments The authors thank Fiorella Foscarini at University of Toronto for fruitful discussions in the analysis of Italian texts, and Sijia Li at Institute of Psychology, Chinese Academy of Sciences for helpful suggestions in the data analysis.
Author Contributions T.Z., J.X., and Y.S. were responsible for study design. X.L., J.C., and W.G. were responsible for data collection. Y.S., P.W., C.C., and T.L. were responsible for data analysis. Y.S., J.X., and T.Z. were responsible for data interpretation. Y.S. wrote the first draft of the manuscript. Y.S., J.X., and T.Z. contributed to the final draft. All authors have read and agreed to the published version of the manuscript.
Funding China Social Science Fund(17AZD041).
Conflicts of Interest The authors declare no conflict of interest.
References