Newsletter N°4: Analysis of online discourse on immigrant and visible minority workers
A project of
En collaboration avec
With the support of
1. Introduction
Discrimination manifests itself as discriminatory social practice, but also as ideology in the form of discriminatory discourse. In effect, by defining and attributing characteristics to an exogenous group (“them”) and an endogenous group (“us”), this ideology legitimizes and normalizes discriminatory practices in society, and it permits these practices to be reproduced. Discriminatory thinking is not innate to human beings, rather it is a learned process and it gets passed down through discourse (van Djik, 2015; Wodak, 2015). Currently, the internet is a location where legitimization and reproduction of the process takes place: according to Hughey and Daniels (2013, p. 342), “the internet is becoming the major means for the production of public opinion and the dominant consensus on ethnic and racial affairs.”
The objective of this study is to analyze online discourse about immigrants and people from visible minorities in the workplace in Quebec. To do this, we collected articles and posts about immigrants or people from visible minorities in the workplace, and we collected comments on these posts, from the Facebook accounts of three major Quebec media outlets (Le Devoir, Journal de Montreal and La Presse) and from two political actors (François Legault and Gabriel Nadeau Dubois) during the period from 2018 to mid-2021. Thanks to a number of tools that we used to collect the data (Eureka, the search tool on Facebook and Facepager), we were able to find 245 posts (including 4 containing no comments) and 63,263 comments (about 14% of which were not used in our analysis, because they contained, for example, only emojis, proper names and/or misspelled words). We analyzed posts and comments in order to paint a portrait of online discourse and to find out how actors communicating differently on social networks impact readers’ reactions to immigrant or visible minority workers in the comments, considering that comments on Facebook are necessarily made reacting to posts on a Facebook page.
Some studies on this topic give us an idea of the amplitude of the discriminatory discourse that exists online against immigrant populations in Quebec. Forcier (2019) found that, in Quebec, approximately three in four online comments about immigrants and racialized minorities are negative, and that the main lines of argument used to define the exogenous group insist on protecting people who make up the majority cultural identity, or immigrants and racialized minorities are presented as an economic or cultural threat or a threat to the legal order (Forcier, 2019). Despite the marked presence of discriminatory remarks online and the literature on hate speech or racist speech online internationally, still very few studies in Quebec focus on this issue.
Analyzing discriminatory discourse online presents several methodological challenges. Indeed, several websites where comments are found (including Facebook) have policies against hate speech and employ tools to automatically detect this type of remark, and many of the accounts of important social actors (including the media and politicians) also moderate comments that appear for their posts. However, all of this does not preclude that discriminatory comments will be present, with the advent of creative techniques that allow one to be discriminatory in less obvious ways, in order to circumvent moderation (Hughey and Daniels, 2013, p. 335-336). These comments, for example, may use coded language with connotation and innuendo. Internet users can also hide behind political rhetoric to make discriminatory remarks, for example, by appealing to “common sense” while basing themselves on collectively shared stereotypes, by characterizing the members of the exogenous group as being catalysts or bearers of problems. They can also play victimization while using vague arguments appealing to “freedom of expression” (Hughey and Daniels, 2013, p. 338-342). These creative techniques make the automatic detection of discriminatory comments a daunting task for researchers hoping to analyze discriminatory speech online.
2. Methodology
To overcome this obstacle, rather than analyzing the discriminatory or non-discriminatory nature of posts and comments directly, we used other means to identify the subjective nature of posts and comments online. By “subjective nature”, we mean the clues left in posts and comments that reveal opinions about immigrant or visible minority workers. Our approach to collecting posts and comments was done in two stages. We collected 245 posts from two politicians (François Legault and Gabriel Nadeau-Dubois) and three major Quebec media outlets (Le Devoir, Journal de Montréal and La Presse) by proceeding first from the search engines Eureka, Facepager, as well as the search engine found on the Facebook site using the following keywords: “refugee”, “immigrant”, “racial”, “racis” (allowing us to search for both “racist” and “racialized”), “stranger” and “arriving”. We then kept only the posts whose main topic was immigrant or visible minority workers. Next, we searched for posts on the Facebook pages of these different media outlets in order to collect the comments for the posts. In this way, we collected a total of nearly 65,000 comments. 14% of these comments were deleted during analysis because they are either emojis, misspelled words or proper names.
Subjectivity in post texts
Types of text: As presented above, we collected posts from three major media outlets and two politicians. The media publish mainly news articles and frequently publish columns, editorials or other types of opinion pieces. Depending on the type of text (politician, article or opinion piece), the media’s treatment of immigrant and visible minority workers is different.
Topics: In recent years, immigrant and minority workers have found themselves at the center of the news on several occasions, and the media’s treatment of this population varies greatly depending on the topic of the text. After collecting the posts, we grouped the posts according to one of the following nine topics:
1) immigrant workers in the healthcare system during the pandemic (guardian angels), 2) difficult working conditions (mainly linked to the working conditions scandal involving migrant agricultural workers during the summer of 2021), 3) discrimination at work, 4) foreign students, 5) economic immigration in general (in connection, for example, with the number of immigrant workers brought in or integration of immigrants into the labor market), 6) law 21 on secularism of the state, 7) immigrants’ work during the pandemic and 8) reform of the Quebec Experience Program (PEQ). At times, the media has also published 9) portraits of specific immigrant cases (an article published in July 2020, for example, describes the lived experience of a Syrian refugee who is now a student and a nurse’s aide).
Position: Another way to examine whether a text has more of a bias than another is to look at the position it takes towards immigrant or visible minority workers on a specific issue. For example, does the text promote an increase or a reduction in the rate of immigration? Or is it proposing an improvement in working conditions or a setback in labor gains? We have classified the texts according to whether they offer solutions or present immigrant or visible minority workers sympathetically, in a neutral manner, or unsympathetically. Although news articles claim to be unbiased, unlike opinion pieces and politicians, they can promote a sympathetic or unsympathetic positions on a political issue when the news item is reported (as is the case, for example, in an article presenting demands issuing from a demonstration or an article presenting a political party’s press release that is a reaction to a policy of the party in power).
Tone: We analyzed the texts of Facebook posts by carrying out a “political tone” analysis which makes it possible to identify the subjective nature of the texts. Political tone is a numerical variable allowing us to identify how much an article or post uses a subjective tone (positive or negative). This variable is obtained thanks to the LSDFr lexicon (created by researchers from Laval University) which classifies how a text is expressed according to the positive or negative political connotation, thus making it possible to determine the overall tone of the text (Duval & Pétry, 2016 ). For example, the word “cure” has a positive connotation while “give up” has a more negative connotation. As Duval and Pétry (2016, p. 200-201) write, “[T]he importance of tone in politics relates to the central role that subjective impressions and emotions play in shaping political attitudes.” This type of analysis is particularly relevant here, since public opinion is sensitive to substantive content, but also to affective content: several studies have demonstrated the influence of affective content (tone) in the media on citizens’ perception of politics (Duval & Pétry, 2016; Miller, 2011; McComas and Shanahan, 1999).
– Popularity (reactions): One way to observe the effect posts have on readers is to look at the number of reactions (also known as “likes” in Facebook parlance) they have.
Comment subjectivity
– Terms used: Several terms are used to describe immigrant and visible minority workers: some have connotations that portray immigrant and visible minority workers positively, and others have connotations that portray them negatively. It is interesting to see when certain connotations show up.
-Sentiments: Similar to political tone analysis, it is possible to carry out an automatic text analysis to detect the subjective character of a text. To do this, we conducted a sentiment analysis of the comments using the FEEL lexicon. This lexicon contains more than 14,000 words and facilitates the analysis of the positive/negative polarity, but also associates certain words with one of the six basic emotions (joy, anger, fear, sadness, surprise and disgust) using Ekman’s emotion typology (Abdaoui et al., 2017). This approach is useful in discourse analysis considering the role that emotions play in politics, for example during rallies, demonstrations or when political actors demand public intervention. Even in the formation of political opinions, emotions affect how we retain information about political topics and act as a guide in how we respond to news (Albertson & Gadarian, 2015; Miller, 2011). This approach differs from political tone analysis since the latter only considers the positive or negative character of words (and not emotions) and specializes in political texts like those written by politicians and the media (unlike FEEL which does not apply to a specific type of text).
-Popularity (reactions): As with posts, one way to see how popular a comment is to look at the number of reactions it has.
3. Results
3.1. Posts and comments by type of actor
Politicians’ Facebook accounts
As previously mentioned, it is not surprising that different actors approach the question immigrants’ or people from visible minorities’ work differently. It is clear that politicians do not claim to be neutral as the media does. On the contrary, their political allegiance is at the heart of their political strategy (and success), and this preoccupation means that there is greater subjectivity in the texts they post. Figure 1 illustrates the distribution of the political tone of the posts and articles of the different actors, both the media and politicians, giving us information about how they deal with immigration issues. This graph is what is called a “boxplot”: the beginning of the box represents the value of the 25th percentile of the political tone of the texts, the line in the middle of the box represents the median, the end of the box represents the 75th percentile and the lines after and before the box (the whisker) represent the range of values. The colored clouds help visualize the data. Among other things, it is interesting to note whether the majority of the texts use a positive, neutral or a negative tone.
Graph 1: Ton politique des texte et publications
We see more polarization between politicians than we do with media outlets when it comes to the tone of the discourse on immigrants or people from visible minorities in the workplace. The Premier of Quebec, François Legault, tends to use a relatively positive tone, often praising his administration’s policies, while MNA Gabriel Nadeau Dubois uses a much more negative tone, as he positions himself in opposition to the current government’s policies. As we can clearly see in Graph 2, which shows the position taken in the posts of the various actors in regards to immigrant workers or workers from visible minorities, this does not mean that François Legault takes political positions that are favorable to these workers, in fact, he is known for having a generally conservative position on social issues, including immigration. During the period studied, his government, among other things, adopted the law on the secularism of the state (Bill 21) and the reform of the Quebec Experience Program, two policies that created more obstacles for religious minorities, in terms of jobs that were available to them, and in terms of areas of study available to foreign students. We see that Legault is the only actor with no texts that favor or defend immigrant workers or workers from visible minorities, while Gabriel Nadeau-Dubois is the only actor who has not posted an unsympathetic text. In short, while in general political tone is a fair measure of a speaker’s appreciation of a topic, in the case of politicians’ Facebook posts, tone is not a reflection of their respective stance on immigration.
Graph 2: Position of actors
As an example, here are two statements published by the aforementioned politicians which demonstrate the relationship between the tone used and the position of politicians’ posts:
François Legault: “What a historic day! The bill on secularism is finally adopted. From now on, religious symbols for state employees in positions of authority will be prohibited. […] I want to salute the hard work of Simon Jolin-Barrette – Member of Parliament for Borduas in this file. He was rigorous and he always kept a respectful tone. Thanks Simon! Tonight, I’m even prouder to be a Quebecer! » Posted on Facebook on June 16, 2019.
Gabriel Nadeau Dubois: “Francois Legault considers immigrants as mere numbers. Today, Ruba Ghazal spoke on behalf of thousands of people who have been harmed by his insensitive rhetoric. Listen to her. And share.” Posted on Facebook May 4, 2021.
This dynamic is felt in the comments on the posts of the two politicians. Indeed, when we compare the feelings conveyed by politician’s comments in Graphs 3 and 4, which show the percentage of the feelings conveyed by the comments, we notice that the main feeling in the comments on François Legault’s posts is proportionally more joy (twice as much as those for Gabriel Nadeau-Dubois), and all of the negative feelings are proportionally less present.
This is evident in the words used: compared to the comments about the media’s posts, comments on Legault’s posts respond directly to the Prime Minister (we see the term “Prime Minister” 6.2 times more often and the term “Mr.” 6.4 times more often) and congratulate him (“bravo” is used 8.5 times more often and “thank you” is used 6.1 times more often). The comments under Nadeau-Dubois’ posts also tend to mention the Prime Minister (which is not surprising since Nadeau-Dubois’ posts challenge him) and also thank Nadeau-Dubois for his work, though less than comments for Legault.
In addition, since the posts are written by politicians, the comments refer more directly to Quebec politics, using the word “politics” 82.85% more often than comments for the media’s posts (with expressions like “political issues”, “immigration policy” and “political party”). Another example of the political dimension of politicians’ posts is notable when we compare Nadeau-Dubois’ posts about an immigrant essential worker threatened with deportation to other texts on the topic of portraits of specific immigrant cases from the media. While newspaper articles carried the stories of these particular immigrant cases as news items, Nadeau-Dubois politicized the issue, and this is noticeable in the comments: we find more words like “politics”, but also “society” (2.7 times more) and “law” (66% more). What’s more, the immigrant’s lived experience is seen more as a struggle, and it is not uncommon to find comments similar to this comment, found under a post dated September 27, 2020: “Thank you for your support to this gentleman who fought for people in your riding.”
Graph 3:
Sentiments of the comments under the publications of Francois Legault
Graph 4:
Sentiments of the comments under the publications of Nadeau-Dubois
Politicians also post less than the media do (they produce a few posts per day rather than a few per hour) and write about a narrower range of topics than the media do (Gabriel Nadeau Dubois has posts about six of the nine topics identified, while François Legault has posts on four of the topics).
In short, compared to media posts, we find a particular dynamic for politicians: because their subjectivity is intrinsic to their work as politicians, the party in power congratulates itself on its good work, while the opposition parties’ critique that work. This leads to more joyous comments for the politician, who congratulates himself and says he is “proud”, and comments making more direct references to Quebec politics.
This particular dynamic is specific to politicians’ posts (and differs depending on the politician’s political allegiance) distinguishing these actors from the media (as is shown in Graphs 1 and 2, where the media accounts have similar results compared to the politicians). This difference means that taking politicians into account in post analysis of the other post variables would create too great a bias. Therefore, the remainder of the analysis in this research report was done solely by taking the media into account.
The media’s Facebook accounts
Does any particular media outlet deal differently with topics related to immigrant and visible minority workers’ work than other media sources? Unlike politicians, the media claims to be impartial, and the way in which the media deals with certain topics can influence readers’ opinions on political issues (McComas and Shanahan, 1999), and even influence elections (Duval & Pétry, 2016). As we saw earlier, in Figure 1, the Journal de Montréal generally uses a more negative political tone in its articles than the other two sources. Another way to examine whether one media outlet is more biased than another is to look at the positions taken in its opinion pieces, as these reflect the outlet’s explicit position. As can be seen in Graph 5 below, which shows the precise position of opinion pieces, since 2018, the Journal de Montréal has also been the media source with the most unsympathetic opinion pieces towards these workers (this is the case for more than 50% of the opinion pieces, a great deal more than in the other media sources).
In essence, when we compare media sources, the two main variables used to measure bias (political tone and position) indicate that the Journal de Montréal presents immigrant and visible minority workers less sympathetically than other media sources. There is, however, no direct correlation in the post’s texts between the political tone and the position taken. The reason is that, as we will see later, tone and position depend on the post’s topic since different ways that workers are depicted impact the tone of the post. Indeed, it is possible, for example, that the texts on a topic are almost exclusively negative, while for other topics, the tone can be as much positive as negative. On the other hand, the reason why texts by the Journal de Montréal are more negative than the other two media sources is not because it publishes more texts on these topics with mostly negative posts. In fact, among the eight topics that are covered by all of the media sources, the Journal de Montréal, on average, has the most negative political tone for four of them (4/8 of the topics, therefore, half) and has unsupportive positions for five topics (5/8). Predictably, the Journal de Montréal is also the medium that has the most positive tone on average for only one of the topics (1/8) and the most sympathetic position for two of the topics (2/8, therefore a quarter). This media source clearly has a different subjectivity (that is more negative towards immigrant or visible minority workers) than the other two media sources.
The subjectivity of a media source can also be measured by the coverage (or lack of coverage) of a specific current topic. It is a fact that media outlets make choices when deciding what information should be news and which should not. A media source that deals with one topic more than other media sources do, for example, shows that it feels that this topic deserves coverage in the public sphere. In this regard, it is interesting to note that the Journal de Montréal did not post any of the 12 articles about the scandal concerning migrant agricultural workers working conditions. Le Devoir published more articles on both topics that involve foreign students (the general topic of foreign students and the reform of the PEQ). In total, 75% of posts (including all opinion pieces) on these two topics were written by this media source alone. The opinion pieces on these topics reflect the media sources’ position: while 26.09% of the articles demonstrate a favorable position towards foreign students, 55.56% of the opinion texts do, and while 43.48% of the articles have a positive political tone, 77.78% of the opinion pieces do. In addition, the comments for the opinion pieces are 27.44% less joyous: among other things, we find the word “thank you” proportionally about 3 times less often (so the comments thank the government less for their policy that negatively impacts foreign students). Debates are also found more in the comments for opinion texts. In fact, the CAQ mentions the verb “to vote” proportionally approximately twice as much, and approximately 9.3 times as much, whether in support of the government or not (for example: “I have complete confidence in the CAQ and if I had to do it again, I would vote for them without a shadow of a doubt”).
Graph 5: Position of opinion pieces
3.2. Posts and comments by topic
One of the most important factors influencing the media’s coverage of immigrant and visible minority workers is the topic of the post. In fact, immigrant and visible minority workers found themselves at the center of the news on several occasions during the period studied, and the media narrative varies greatly depending on the occasion. Figure 6 shows the political tone of the articles by grouping the topics into nine categories: 1) immigrant workers in the health system during the pandemic (guardian angels), 2) difficult working conditions (mainly related to the working conditions scandal surrounding migrant agricultural workers in the summer of 2021), 3)discrimination at work, 4) foreign students, 5) economic immigration in general (related, for example, to the number of immigrant workers brought in or integration of immigrants into the job market), 6) Bill 21 on secularism of the state, 7) immigrants work during the pandemic, 8) reform of the Quebec Experience Program (PEQ) and 9) portraits of specific immigrant cases. We see that while the majority of topics have just as many posts that use a positive tone as those that use a negative tone, posts on some topics (including the pandemic’s effect on immigrant workers, workplace discrimination, migrant agricultural workers working conditions) overwhelmingly use a more negative political tone.
Graph 6: Political tone of online texts and publications
Moreover, when we look at positions taken in posts according to topic in Graph 7, we see that the topics fall into two categories: topics for which the majority of posts are sympathetic to migrant and visible minority workers (discrimination in the workplace, migrant agricultural workers working conditions, portraits of particular cases of immigrants and guardian angels (immigrant workers in the health sector during the pandemic)) and other less sympathetic and more neutral ones, with Bill 21, on the secularism of the State, standing out as having proportionally the most unsympathetic posts about workers from religious minorities.
Graph 7: Positioning of online publications and texts
We can see that, although political tone is a measurement that allows us to analyze subjectivity in a political text, a negative political tone is not always synonymous with an unsympathetic position. In fact, the texts on the topic of Bill 21, for example, although proportionally more unsympathetic, do not use a particularly negative political tone. On the contrary, the texts on the topics of working conditions and discrimination, for example, are mostly sympathetic to immigrant and visible minority workers, but use a predominantly negative political tone. The reason is that, like Gabriel Nadeau-Dubois’ posts, these posts defend immigrant and visible minority workers (victims of discrimination and enduring poor working conditions) by describing discrimination in their working conditions negatively. Therefore, we note that for these two topics, there is a more flagrant media bias. For example, in an article published by the Journal de Montréal on May 4, 2019, detailing discrimination experienced by hockey players: “Hockey player Jonathan-Ismaël Diaby, targeted by racist insults last week during a LNAH game in Saint-Jérôme, was on the set of Tout le Monde en Parle on Sunday. He recounted the sad events and discussed possible solutions alongside Georges Laraque, who was also bullied on ice rinks as a youth […].”
Table 1: Terms used to describe migrant and racialized workers by story topic (%)
The terms used in comments to refer to immigrant or visible minority workers also vary according to the topic of the post. Some terms present the workers as a benefit to society (guardian angels) or evoke sympathy (asylum seekers, refugees) while other terms present them as a social nuisance (unemployed). In Table 1, we can see the recurrence of terms used in the comment sections to refer to immigrant and visible minority workers. The columns represent different terms, and the rows indicate the topic of the article where the comment was found (the bottom row refers to the general percentage of terms used). The topics that are overrepresented for each term (at least 2 times more present) are in green and those that are less so (at least 2 times less present) are in red. For example, although 2.42% of the total terms collected that refer to immigrant or visible minority workers as “asylum seeker”, we see that for the topic of workers in the health system during the pandemic, “asylum seekers” represents 18.63% of the terms. Some results are not surprising (such as “guardian angel” to designate workers in the health system during the pandemic, or “religious signs” to designate those affected by Bill 21), but it is interesting to note that the more positive terms in the comments for posts are associated with the topics of workers in the health system during the pandemic and portraits of specific immigrant cases (both of which have many proportionally more sympathetic posts) and that the term “unemployed” is used more in post comments about the effect of the pandemic on migrant workers (a topic with fewer supportive posts and a more negative political tone).
Some topics are also more popular (i.e. have more reactions) than others. We can see, in Table 2, the average number of reactions to posts, the median number of reactions to posts, and the percentage of posts with more reactions than the overall median. Foreign students was one of the topics whose posts generally receive little reaction (most of the posts on this topic, as mentioned, are written by Le Devoir), in addition, the topics of discrimination at work and migrant workers working conditions, are topics where the posts have a negative tone and present these workers as experiencing injustices. Furthermore, two topics stand out for the generally higher number of reactions they elicit for posts: Bill 21 and portraits of specific immigrant cases. These two topics are representative of two ways of depicting immigrant and visible minority workers: several posts about Bill 21 portray workers who are religious minorities in the public system as a threat to Quebec society (promoting values contrary to Quebec values) and several posts are portraits of specific immigrant cases describing the difficulties of their lived experiences or the demanding work they do in Quebec, thereby arousing sympathy and admiration.
Table 2: Popularity (reactions) of the subjects according to the media
This difference in representation is noticeable in the comments for the posts. In particular, for the topic of special cases, we notice 50.91% more comments connected to the feeling of joy than for those for the topic of Bill 21 (with greater presence of words such as “thank you”, “welcome”, “dream” and “welcome”) and fewer comments connected with anger and fear. Figures 8 and 9 display word clouds where we can see words that are overrepresented in the comments for the topic (that is, these words are found at least twice as often in the comments), the words that are larger are used more often. We find that when it comes to Bill 21, the expressions that come up most frequently are abstract and general concepts (religion, women’s rights, freedom, law, authority), whereas when it comes to of the portrait of an immigrant, our data shows that the words used are much more humanizing and more “down to earth” (thank you, beautiful, need, family, help). These topics are representative of the trends observed: the topics where there are posts that are more sympathetic towards immigrant and visible minority workers (discrimination in the workplace, migrant agricultural workers working conditions, portraits of particular cases and workers in the health system during the pandemic) portray immigrants as being a benefit to society (for example, the guardian angels) or as victims of inequality (for example, workers with poor working conditions), and humanize these workers. On the other hand, other topics such as Bill 21 and the effect of the pandemic on immigrant workers give rise to comments that refer to more general concepts and depict immigrant or visible minority workers as a nuisance. In the case of the topic of the effect the pandemic had on migrant workers, not only do the comments contain the word “unemployed” to describe these immigrants 34.42 times more frequently than the other topics, but, in addition, the comments are generally less related to joy and more related to sadness, fear and disgust. Certainly, given the pandemic, there has been an intensification of discriminatory practices and discourse (especially towards the Asian community) since COVID-19 and its variants come from outside of the country and are often associated with foreigners arriving by plane. This intensification is particularly observable on social networks, where there are more racist comments than before the pandemic (Dubey, 2020).
Graph 8:
Words overrepresented in
Comments on Bill 21
Graph 9:
Words overrepresented in
comments on the portrait of particular cases of immigrants
Case study: migrant agricultural workers
Immigrant and racialized visible minority workers comprise a large population that includes people who do different types of work and who have different immigrant profiles: people affected by the PEQ reform, for example, do not have the same profile as workers in the health care system during the pandemic (generally speaking, of course, because it is quite possible that an immigrant working in the health care system is also a student using the PEQ). A good way to see how different post topics change the way immigrants are depicted is to follow a specific immigrant profile and see how that population is represented differently. This is the case, for example, of migrant agricultural workers who have made headlines in the global topic of economic immigration, the scandal over migrant agricultural workers’ working conditions and the pandemic’s effects on the immigration of agricultural workers. Here are examples of article headlines about migrant agricultural workers for each of the topics:
- Economic immigration: “Our agriculture survives thanks to Latino workers” Journal de Montréal, published on Facebook on August 10, 2019
- Working conditions: “Unsanitary housing for workers – The president of Demers apologizes” La Presse, published on Facebook on June 4, 2021
- Effects of the pandemic: “Forty foreign agricultural workers have been infected in Quebec” La Presse, published on Facebook on June 24, 2020
Among media posts we collected on economic immigration in general, four of them focus specifically on the agricultural sector’s dependence on migrant agricultural workers, thereby depicting agricultural workers as beneficial for Quebec society. These four posts have, on average, 3 times more reactions than the other two topics and use a more positive tone. Comparatively, we also find about 71.71% more comments related to joy, using a greater proportion of words such as “thank you” and “contribute.”
Several articles about immigrant workers’ working conditions focus on the long working hours, harsh conditions and unhealthy living conditions of migrant agricultural workers, thereby depicting these workers as victims of injustice. We find the most posts that are sympathetic to migrant workers on this particular topic in addition to a more negative political tone. There are also 56.07% more comments related to anger that use a greater proportion of words such as “right”, “shameful” and “inhumane”.
Finally, seven posts about the effects of the pandemic focus specifically on migrant agricultural workers (for example on the challenges of bringing these workers in from abroad). None of the posts take an adversarial position towards these workers, but compared to the other two topics, more texts take a neutral position and the political tone used is predominantly negative compared to the general topic of economic immigration. In the comments, we note that migrant agricultural workers are sometimes seen as a nuisance, as people coming to Quebec to take advantage of our system: this is the only topic where the term “unemployed” is used to refer to these workers and certain words such as “dangerous” and “lazy” are used proportionally more.
In short, the way the media covers migrant agricultural workers (and the reaction to this coverage in the comments) clearly shows how an immigrant population can at times be seen as a benefit to society, sometimes as victims and sometimes even as a danger, depending on the angle from which the media approach it, as well as the topic of the article.
3.3. Posts and comments according to position and political tone
Effect of position
Negative positions taken by media sources are generally more popular (i.e. they have more reactions and comments) than positive positions, and this goes for the vast majority of topics (although this should be taken with a grain of salt for certain topics where there are few unfavorable articles on certain topics). We saw earlier that the Journal de Montréal tends to have more unfavorable texts than the other two media, and this tendency for unfavorable texts to have more reactions is perhaps a cause of this, considering that the Journal de Montréal is the Facebook page among the three media outlets with the most subscribers. However, this trend holds true regardless of the media outlet, as can be seen in Table 3, whether it is the Journal de Montréal, La Presse or Le Devoir, posts with an unfavorable position have more reactions and comments than favorable posts. The difference, on the other hand, is especially pronounced for the Journal de Montréal (and less for Le Devoir, whose favorable posts have, on average, more reactions, but fewer median reactions and fewer comments). This trend also holds if we look at opinion pieces and articles independently. This does not necessarily mean that more people are against the plight of immigrant or visible minority workers; reactions on Facebook can range from ‘likes’ to ‘haha’ or ‘angry’ (which someone might indicate if they find a position ridiculous or disagree with the post), and the comments may disagree with the statement reported by the post. This greater popularity of negative positions is still an important phenomenon, especially considering that the media selected are companies working to make a profit, so their goal is to have as many reactions as possible on their posts, in order to be more popular.
Table 3: Popularity of opinions according to the media
Effect of Political Tone
As we saw earlier, posts on some topics tend to have a more negative political tone (namely the topics of discrimination and working conditions) but have a positive stance in regard to immigrant or visible minority workers. This leads to the workers being portrayed as victims who are suffering injustices. Posts that portray immigrant or visible minority workers as beneficial to society tend to use a more positive political tone, and this is noticeable when comparing, for the topic of the portrait of specific cases, posts that have a positive tone and those that have a negative tone. In fact, this topic includes both texts that highlight the difficult lived experiences that the particular person in the case had to undergo (thus portraying the person in question as a victim and using a more negative tone), and others highlighting the successful integration of a person in a particular case (therefore portraying that person as a benefit to society and using a more positive tone), and some even emphasize both. As we see in Graphs 10 and 11 which show the distribution of feelings in the comments for posts on this topic according to the tone used, different political tones mean different feelings in the comments: posts using a negative political tone have comments evoking more negative feelings like fear or disgust, and comments describe the situation of the particular case more as “damaging”, and talk about “problems” in the “system” or “law”. Posts using a positive political tone get comments that evoke more surprise and joy, in addition to using more words saying, “thank you” for their “gesture”, “welcome”, or highlight the “contribution” of these particular individuals. Here are excerpts from posts that clearly show the difference between texts with a negative political tone portraying immigrant or visible minority workers as victims, and texts with a positive political tone portraying immigrant and visible minority workers as benefits to society.
Graph 10
Sentiments of comments from negative publications about the portrait of individual cases
Graph 11:
Sentiments of comments from positive publications about the portrait of individual cases
-A negative political tone in relation to the profile of specific cases: “Guatemalan temporary workers claiming to be victims of fraud who challenged their deportation in Federal Court lost their case. […] ” We were modern slaves for Canada,” said Juan Godoy over the phone, his voice breaking up.” Le Devoir, published on Facebook on December 8, 2018.
-A positive political tone in relation to the profile of specific cases: “A generous refugee barber from Rwanda will cut hair for people who cannot afford this luxury for free, every last Monday of the month, for almost a year to give back to the community » Le Journal de Montréal, published on Facebook on October 19, 2019.
Effect of position and tone on comments
Surprisingly, tone and position have a minimal effect on comments, except where different positions or a different political tone for the same topic may be associated with different ways that immigrant or visible minority workers are characterized, as is the case for the topic “portrait of particular cases of immigrants”. In fact, as we see in Table 4, which presents the tone used, the number of reactions and the feelings expressed in the comments for articles on Bill 21, the feelings we see in the comments for posts that are negative towards these workers are not very different from comments for posts with positive positions. Admittedly, there is a difference (about 8.44% of the comments have happiness as their main feeling compared to 6.60% of the comments for positive posts), but this is a relatively small difference. Analysis of the terms used in the comments is also inconclusive. This does not necessarily mean that the tone and position of a post does not influence comments: it is quite likely that there is a difference, however, the types of tools chosen for the analysis chosen are not the best to detect this.
Table 4: Articles on Bill 21
4. Conclusion
The automatic analysis of the subjectivity of Facebook posts and comments by Quebec social actors (media and politicians) relating to immigrant workers or workers from visible minorities allows us to paint a portrait of the online discourse about this population, in particular to identify different media strategies as well as their effects in how this discourse is received. Politicians’ posts are notably very different from media posts since politicians’ success is intrinsically linked to their subjectivity, which means that their posts are considerably more subjective (both in tone and in position) than the media’s posts, which on the contrary claim journalistic neutrality. In addition, comments tend to praise politicians and further politicize immigration issues.
Media outlets, although claiming journalistic neutrality, are not entirely neutral. Whether through opinion pieces, comments reported in articles or in the political tone adopted, subjectivity is detectable in the media. Le Journal de Montréal, for example, promotes more positions that are unfavorable to immigrant workers and visible minority workers and uses a more negative tone in its texts than Le Devoir and La Presse. Also, posts that promote a negative position usually get more reactions than those promoting a positive position. This may partly explain why Le Journal de Montréal is also the medium with the most reactions to their posts, on average. The media are for-profit companies looking to garner the most reactions to their posts on social networks, and the Journal de Montréal stands out from the others because it promotes negative position towards immigrant or visible minority workers. The media’s subjectivity is also noticeable in the events it covers, Le Journal de Montréal opting not to cover the scandal involving the working conditions of migrant agricultural workers in the summer of 2021 or Le Devoir covering more issues relating to foreign students.
Immigrant and visible minority workers have made headlines on different occasions since 2018, and media posts had a significant impact on how these workers were presented. Depending on the topic, three different ways of representing immigrant workers or visible minority workers stand out: either they are seen as victims of an injustice, as a benefit to Quebec society or even as a danger to Quebec society. For example, Bill 21 and the pandemic have been occasions where these workers have been portrayed as a danger to Quebec and its values, or as outsiders taking advantage of Quebec society both in media posts and in comments. These statements are discriminatory since they serve to justify the exclusion of an exogenous group from the “we” (that is, Quebeckers). The comments characterizing these workers as being a danger refer to lofty principles for justification, while the other two depictions humanize immigrant and visible minority workers and are more “down to earth”. There is also a difference between characterizing these workers as victims and presenting them as beneficial, the first characterization arouses more negative feelings like anger in comments, and the second characterization arouses positive feelings such as joy.
The analysis on the subjective nature of posts and comments, on the other hand, was not a perfect substitute for assessing whether comments were discriminatory, and this is especially true for politicians’ posts, which are overtly subjective. In fact, comments characterizing immigrant workers or visible minority workers as dangers do not stand out markedly from other comments when a feeling analysis is carried out, unlike the two other depictions where it is possible to associate feelings like joy or anger. In addition, while it is possible to qualitatively identify positions in posts, it is impossible to identify positions in comments considering the staggering number of comments collected. Take the example of the following comment written under a post by François Legault on November 5, 2019, about foreign students: “Bravo, the law is the law, give them an inch they’ll take twelve.” This is decidedly discriminatory since it gives a negative attribute (taking advantage of the benevolence of Quebecers) to an exogenous group, but at the same time it is difficult to detect it automatically considering that it does not clearly define the endogenous group (it is a rather vague “they”), cannot be attached to a negative feeling or tone (on the contrary, we even find “bravo” which is rather positive) and does not contain an insult nor a racist term. A more in-depth study of discriminatory discourse on social networks will necessarily require the development of a machine learning algorithm capable of differentiating discriminatory comments from other comments.
5. Bibliography
Albertson, B., & Gadarian, S. K., (2015), Anxious Politics : Democratic Citizenship in a Threatening World, Cambridge University Press, 268 pages
Dubey, A. D., (2020), The Resurgence of Cyber Racism During the COVID-19 Pandemic and its Aftereffects : Analysis of Sentiments and Emotions in Tweets, JMIR Public Health and Surveillance, vol. 6, no. 4, 7 pages
Duval, D., & Pétry, F., (2016), L’analyse automatisée du ton médiatique : Construction et utilisation de la version française du Lexicoder Sentiment Dictionary, Canadian Journal of Political Science/Revue Canadienne de Science Politique, vol. 49, no. 2, pages 197‑220
Forcier, M., (2019), La construction des frontières nationales à l’ère numérique : Analyse critique des discours en ligne sur l’immigration et les minorités racialisées au Québec, Thèse de doctorat, Université de Montréal, 488 pages
Hughey, M. W., & Daniels, J., (2013), Racist comments at online news sites : A methodological dilemma for discourse analysis, Media, Culture & Society, vol. 35, no.3, pages 332‑347
Kreis, R., (2017), #refugeesnotwelcome : Anti-refugee discourse on Twitter, Discourse & Communication, vol.11, no.5, pages 498‑514
McComas, K., & Shanahan, J. (1999), Telling Stories About Global Climate Change: Measuring the Impact of Narratives on Issue Cycles, Communication Resaearch, vol. 26, pages 30-77
Miller, P. R., (2011), The Emotional Citizen : Emotion as a Function of Political Sophistication, Political Psychology, vol. 32, no.4, pages 575‑600.
Passerello, G. L., Hazelwood, J. E., & Lawrie, S., (2019), Using Twitter to assess attitudes to schizophrenia and psychosis, BJPsych Bulletin, vol.43, no.4, pages 158‑166.
Wodak, R., (2015), Discrimination via Discourse, The Routledge Handbook of Linguistic Anthropology, pages 366
