This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Local and organic foods have shown increased importance and market size in recent years. However, attitudes, sentiment, and habits related to such foods in the context of video social networks have not been thoroughly researched. Given that such media have become some of the most important venues of internet traffic, it is relevant to investigate how sustainable food is communicated through such video social networks.
This study aimed to explore the diffusion paths of local and organic foods on YouTube, providing a review of trends, coincidences, and differences among video discourses.
A combined methodology involving webometric, framing, semantic, and sentiment analyses was employed.
We reported the results for the following two groups: organic and local organic videos. Although the content of 923 videos mostly included the “Good Mother” (organic and local organic: 282/808, 34.9% and 311/866, 35.9%, respectively), “Natural Goodness” (220/808, 27.2% and 253/866, 29.2%), and “Undermining of Foundations” (153/808, 18.9% and 180/866, 20.7%) frames, organic videos were more framed in terms of “Frankenstein” food (organic and local organic: 68/808, 8.4% and 27/866, 3.1%, respectively), with genetically modified organisms being a frequent topic among the comments. Organic videos (N=448) were better connected in terms of network metrics than local organic videos (N=475), which were slightly more framed regarding “Responsibility” (organic and local organic: 42/808, 5.1% and 57/866, 6.5%, respectively) and expressed more positive sentiment (M ranks for organic and local organic were 521.2 and 564.54, respectively, Z=2.15,
The results suggest that viewers considered sustainable food as part of a complex system and in a positive light and that food framed as artificial and dangerous sometimes functions as a counterpoint to promote organic food.
Sustainability has been receiving global attention with the United Nations promotion of Sustainable Development Goals. Goal 2 “Zero Hunger” involves food security and agriculture development [
The global organic food market size was estimated at US $90 billion in 2016, with the United States, Germany, and France being the main consumers and India, Uganda, and Mexico being the main producers [
On the other hand, the marketing of food products showed a change in the 2000s from industrialized processes to artisanal, small, and locally based processes [
There is evidence that sensationalist and erroneous content is being fueled by social media search engines owing to their business focus [
To find diffusion paths of local and organic food products on YouTube by collecting information on related videos and comparing their network levels with social network analysis.
To review trends and differences among discourses through framing analysis on video content.
To explore the opinions, attitudes, behaviors, and emotions expressed by viewers through semantic and sentiment analyses on comments extracted from the videos.
With the advent of internet 2.0, online collaboration and activism increased, transforming the internet into a conversational space through the rise of social networking sites. Internet 3.0 incorporated location and real-time aware devices and apps, prompting more personalization of products and services.
Social networks can reduce the communication gap between producers, consumers, and other interested people. For example, business engagement on Twitter is related to consumers’ web-based word-of-mouth communication, and its influence reaches consumers with a second-degree relationship to brands [
Previous social media studies on food communication through text included two-way communication by public organizations related to food safety and nutrition, and it was found that the main themes were queries and complaints, benefits of social media in query and complaint services, content redesign driven by social media use, and social media to learn about consumers [
Structural and social factors of web-based communication channels affect their roles in the image construction process of organic food brands [
Comments from Mexico’s Starbucks Facebook page (a shop chain that sells organic coffee) reflected that people interacted more through happiness, but anger and longing were often used to generate participation [
However, video networks related to organic or local food have not been thoroughly researched. Video social networks account for over half of the internet traffic when measured in bytes [
There is no consensus on what is organic or local food. A definition adapted from previous reports [
With regard to local food, the most recognized feature is the marketing arrangement [
According to applicable institutional and national guidelines and regulations, ethics approval was not required for this study, as we focused on publicly available YouTube data. Video data were extracted in 2015 with YouTube Data Tools [
The following three queries were used to extract video data: “organic food,” “local food,” and “local organic food.” The resulting three files were appended into one. Videos that appeared only with the “organic food” query were labelled as “organic food videos.” There were no videos that appeared exclusively with the “local food” query; thus, all videos that were not labelled as “organic food videos” were labelled as “local organic food videos.” Videos were watched by an investigator, and in case the content was not in an understandable language, the investigator requested a native speaker to interpret the content. Content that clearly was unrelated to food was discarded, reducing a total of 964 videos to 923 videos. The videos were also classified according to country and uploader. The types of uploaders considered were as follows:
Business: It included businesses related to food production, processing, and distribution. Businesses linked to health, tourism, and banking were added to this category as well.
Community: It included citizens and communities.
Education: It included citizens disclosed as professors, researchers, students, and lecturers; research institutions; universities; and informal education-related accounts.
Media: It included both traditional and internet-based media.
Others: It included government, politicians, and celebrities.
Undisclosed: It included all accounts that did not fit in the previous categories.
Video content was also classified according to food frames. Framing is the action of using images and words to influence how audiences interpret a message, promoting specific versions of reality. They are useful tools to infer what people think is important.
Food-related framing usually falls into the corporative-political category [
Condensed frame packages for sustainable food products.
Frame | Emotional basis | Key concepts | Visual cues | Textual cues |
Responsibility | Endearment | Accountability and vulnerability | Children, fragile plants, and young animals | Caring, future generations, our children, and vulnerability |
Undermining of foundations | Alarm and concern | Balance, base, complex systems, and links | Interconnections between elements in the ecosystem | Fragile balance, mutual dependency, and unstable |
Frankenstein | Anxiety and unscrupulousness | Apocalypse, Pandora’s box, and sorcerer’s apprentice | Monsters and skulls | Frankenstein food, poison, and risks |
Natural goodness | Admiration and astonishment | Authenticity, good taste, health, and purity | Idyllic nature and products | Natural, pure, and taste |
Progress | Trust | Modernization and progress | High-tech tools | A better world, constant striving, and technology |
Good mother | Gratitude, enjoyment, and love | Freedom of choice and great variety of products | Pleasure of shopping and rich harvest | Friendliness and product range |
This type of analysis involves theories, methods, and techniques to study social relations and their structures [
Based on the study by Tsou et al [
This is a type of network study where the unit of analysis is keywords. The software employed was TI, an open source tool that generates a word frequency list, a word-occurrence matrix, a word co-occurrence matrix, a normalized co-occurrence matrix, and a word list from a set of short texts [
The sentiment analysis aimed to determine the polarity of text through natural language processing. Although most sentiment studies on social networks do not consider emoticons, this tendency has reverted in recent years, as their inclusion increases accuracy. Thus, the value of emoticons (positive, neutral, or negative) was assigned based on the SentiStrength software package [
Among the 923 videos related to organic and local food, 448 were included in the organic food video list and 475 in the local organic food video list. Overall, 606 videos disclosed location (47 countries). As the keywords employed were in English, the lists contained videos mostly from the United States (n=393). The second most frequent location was undisclosed (n=317), followed by India (n=25) for organic food videos and Canada (n=39) for local organic food videos. Overall, organic food videos had higher metrics in terms of views, likes, dislikes, and comments (
As for types of uploaders, media-related YouTube channels were the most common for both lists (
Nonparametric test of YouTube metrics.
Metric and group | N | Mean rank | Sum of ranks | Z | |
|
923 |
|
|
11.056a | |
|
Organic | 448 | 561.90 | 251729.00 |
|
|
Local organic | 475 | 367.78 | 174697.00 |
|
|
905 |
|
|
12.655a | |
|
Organic | 441 | 565.08 | 249199.50 |
|
|
Local organic | 464 | 346.48 | 160765.50 |
|
|
905 |
|
|
8.319a | |
|
Organic | 441 | 515.27 | 227234.00 |
|
|
Local organic | 464 | 393.82 | 182731.00 |
|
|
899 |
|
|
9.304a | |
|
Organic | 436 | 528.26 | 230323.00 |
|
|
Local organic | 463 | 376.30 | 174227.00 |
|
a
Types of uploaders of organic and local organic food videos.
Uploader | Organica (N=448) | Local organica (N=475) |
Business | 37 | 87 |
Community | 78 | 94 |
Education | 93 | 74 |
Media | 194 | 157 |
Others | 7 | 17 |
Undisclosed | 39 | 46 |
aχ21 (N=923)=4.15;
The top organic-related videos in terms of views and likes were uploaded mostly by media channels and contained short educative facts about food products. However, some of them were sensationalist. The top video was “Grocery Store Wars,” a stop-motion parody of the movie franchise Star Wars, contrasting organic food products and conventional food products in a supermarket. On the other hand, the top videos related to local organic food were business oriented and sometimes employed humor. The video with the highest number of views and likes was a three-dimensional animation commercial, which was part of a campaign by Chipotle Mexican Grill, an American restaurant chain. This video contrasted chemically treated and mechanically processed food products with food produced and processed by farmers. It can be concluded that the top videos in both cases were story-telling driven and showed a contrast between sustainable and nonsustainable foods.
Based on van Gorp and van der Goot [
Types of frames in organic and local organic food videos.
Frame | Organica (N=808), n (48.26%) | Local organica (N=866), n (51.73%) |
Good mother | 282 (34.9%) | 311 (35.9%) |
Natural goodness | 220 (27.2%) | 253 (29.2%) |
Undermining of foundations | 153 (18.9%) | 180 (20.7%) |
Frankenstein | 68 (8.4%) | 27 (3.1%) |
Responsibility | 42 (5.1%) | 57 (6.5%) |
Progress | 43 (5.3%) | 38 (4.3%) |
aχ21 (N=1674/923)=4.84;
In order to visualize how different is the structure of the two video groups, their network metrics were compared. Because both lists share ties, metrics for the entire video network were also calculated. Local organic food videos had higher connected components, modularity, diameter, and average path lengths (
Network centralities.
Centrality name | Description | Organic network | Local organic network | Organic and local organic network |
Weakly connected components | Subgroups of nodes that can be reached from every other node in the group. | 48 | 155 | 197 |
Density | Total number of ties divided by the number of all possible ties that can exist within a network. | 0.038 | 0.005 | 0.007 |
Modularity | The strength of the division between subgroups in a network. | 0.197 | 0.471 | 0.329 |
Diameter | Average of the maximum distance between the nodes of a network. | 12 | 16 | 15 |
Path length | Average of the distance between the nodes of a network | 3.902 | 4.640 | 4.652 |
Number of nodes | N/Aa | 448 | 475 | 923 |
Number of shortest paths | N/A | 95,908 | 28,967 | 263,249 |
Degree | Average number of direct connections a node has to other nodes. | 9.67 | 2.50 | 6.75 |
Clustering Coefficient | Measure of how close a node is to be part of a group. | 0.191 | 0.081 | 0.142 |
Closeness | Average number of steps to access all the other nodes in a network. | 3.272 | 2.509 | 2.287 |
Betweenness | Number of shortest paths that connect other nodes in the network by passing through a specific node. | 621.457 | 222.01 | 532.013 |
aN/A: not applicable.
The local organic food video network and the organic food video network tied together. The organic food video network is presented as a star with central videos reaching the most distant videos within the network.
The top videos in terms of betweenness and in-degree centrality explained the basics of organic and local food and were predominantly uploaded by media channels. In contrast, videos with high out-degree centrality were uploaded mostly by businesses and individuals. A few organic-related videos identified the food as expensive, while local organic-related videos usually presented a specific area where such food products were available.
Spearman correlation analyses between YouTube metrics and network-related metrics were performed to find how much the popularity features are related to communication patterns in the network. The number of views, likes, dislikes, and comments were moderately correlated with degree, modularity class, eigenvector, and betweenness centralities in the organic food video network (
Spearman correlations for the organic network.
Variable | Degree | Modularity class | Clustering coefficient | Betweenness |
Views | 0.209a | 0.174a | −0.185a | 0.218a |
Likes | 0.187a | 0.192a | −0.201a | 0.182a |
Dislikes | 0.240a | 0.120 ( |
−0.180a | 0.193a |
Comments | 0.182a | 0.157 ( |
0.191a | 0.164 ( |
a
Spearman correlations for the local organic network.
Variable | Degree | Modularity class | Clustering coefficient | Betweenness |
Views | 0.156 ( |
0.116 ( |
0.015 ( |
0.287a |
Likes | 0.087 ( |
0.075 ( |
−0.030 ( |
0.239a |
Dislikes | 0.086 ( |
0.039 ( |
0.029 ( |
0.177a |
Comments | 0.143 ( |
0.081 ( |
0.045 ( |
0.257a |
a
The Gephi software was used to visualize the semantic networks corresponding to the two video lists. The 107 most frequent words found in the comments sample from the organic food videos were represented with nodes (
The word “organic” was the most frequent in these comments, with the term “food” closely related to it. Verbs connected to “organic” were “grow,” “know,” “like,” “need,” “say,” and “think.” Frame-related words are presented in
Semantic network for organic food video comments. The size reflects the word frequency. Ties show which words were found in the same comment, with tie thickness reflecting the frequency of such relationships.
Semantic network for local organic food video comments.
In summary, while both groups of comments featured fruits and vegetables, organic food videos featured dairy and proteins, and local organic food videos featured grains and cereals. Moreover, the organic food network had a greater variety of Frankenstein frame–related words.
There was no relevance with regard to negativity valence. However, comments made on local organic videos had a higher rank for positive valence in comparison with comments made on organic videos (
Nonparametric test: sentiment valence of comments made on the videos.
Valence and group | Mean rank | Sum of ranks | Z |
|
|
|
|
|
2.159 | .03a | |
|
Organic | 521.2 | 403927 |
|
|
|
Local organic | 564.54 | 163718 |
|
|
|
|
|
0.15 | .88a | |
|
Organic | 532.27 | 4125409 |
|
|
|
Local organic | 534.95 | 155136 |
|
|
aN1=775, N2=290.
Organic and local food products were communicated on YouTube by several actors articulating efforts to educate the public about food. In particular, the position of organic food videos in the network strengthened the diffusion paths, with a better structural capital than local organic videos, whereas the high in-degree (ties directed to a video) suggested a broadcast network. The relationship between network metrics and negative reactions (number of dislikes and less positive sentiment in comments) implied that negativity might play a relevant role in the diffusion of videos. This coincides with the theory that negative relationships might explain social network outcomes better than positive relationships [
Another relevant finding was the words related to “organic” and “local.” The frames have been summarized in
Words like “love” and “thanks” were used often in the case of comments for local organic videos. As they also employed the Responsibility frame more frequently, it implies a more human and social dimension for the word “local.” It also has implications for activating the viewer in areas besides food consumption, as images involving future generations and nonhuman living beings are part of this frame. This partly explains why comments for local organic videos were more positive than those for organic videos.
Finucane et al [
The persistence of uncomfortable feelings among the public towards GMOs, which are considered as “monsters” according to the Frankenstein frame, points to a failure in scientific communication. This is particularly true regarding information on pesticides, soil depletion, and nutrition. Such communication patterns could be improved by closing the gap between scientists and the public, as attempted by some of the videos uploaded by education channels. This could potentially bring more transparency and trust to the food production chain.
Dissimilar ideas of what is food influence the outcomes of guidelines and food policies. There is little knowledge on the effect of YouTube’s ranking criteria in areas other than politics and entertainment. Hence, a multimethod analysis of the communication of basic human needs, such as food, can provide us with more understanding of the consequences of algorithm usage in rich social media and its interplay with human users. Moreover, we uncovered communication patterns and specific visual and textual cues, providing more comprehensive insights of a complex ecosystem of actors involved in food production, distribution, and consumption. The case of organic and local food intertwines consumerism with environmental concerns that have the potential to impact public health.
As the language of analysis was English, there was partially limited access to videos from non-English speaking cultures. Sentiment can be influenced by many contextual features, such as weather [
This study explored the communication paths, discourses, opinions, attitudes, behaviors, and emotions related to sustainable food products in YouTube video networks. Based on our objectives, we can make the following conclusions:
The organic video network was more consolidated than the local organic video network and was driven mostly by media and business YouTube channels.
The “Good Mother” and “Natural Goodness” frames were most frequently employed in the videos, followed by “Undermining of Foundations.”
The concept “organic” has become consolidated among both specialists and the public, while the term “local” is in the process of acquiring a formal definition. Nevertheless, the term “organic” was slightly more associated with health risks and negative feelings, while the term “local” was perceived as more human/social and more positive.
Further studies could incorporate more languages, as well as a larger data set. Time-based analysis and segmentation of semantic analysis across different geographical locations would also deepen the present understanding of the diffusion of both organic and local food products. Another area for further exploration is GMOs, as this technology will continue to transform food production and consumption patterns in contrast with traditional agricultural methods.
application program interface
genetically modified organism
The authors are grateful to Deepak Vijay, Priyanka Ram, Reach Syden, and Sittiphan Jiyavorananda for their help with translation. A previous version of this article was presented at the 6th International Kansei Engineering and Emotion Research Conference, UK in 2016.
None declared.