Abstract

With the rapid development of China’s S&T, public discourse is constantly reshaped by internet buzzwords which have become the digital carriers of social mentality and the emotion of the time. This research applied the internet buzzwords from 2000 to 2026 as its object, covering the different stages of Chinese internet era. Based on mainstream Chinese net-media, 20 representative internet buzzwords were collected annually, yielding 540 entries. After data screening of synonym merging, cross-year deduplication, and integration of etymologically related semantic items, 18 redundant duplicates were removed, resulting in a final analytical database comprising 522 valid internet buzzwords. Integrating the HowNet Emotional Dictionary, DUTIR and Chinese Emotional Lexicon, a five-tier emotional valence annotation system was constructed, categorizing the items into Strongly Negative (N1), Weakly Negative (N2), Neutral (N3), Weakly Positive (P4), and Strongly Positive (P5). In conjunction with linguistic features, specialized manual annotation rules were developed and conducted with stratified quantitative statistics of emotional distribution across the full sample. The results shew that the corpus comprehensively covered five major content domains varying from public expressions, social headlines, policy discourse, youth subculture, to industrial hotwords, and features long temporal coverage, all-scenario representation and stratified annotation. Weakly Positive buzzwords reached 247, accounting for the largest proportion and serving as the primary emotional carrier for daily entertainment and aesthetic expressions among netizens. Neutral items numbered 221, increasing year by year with the development of digital industries and AI technologies. Weakly Negative expressions came to 176, centrally carrying the life pressures and self-relieving emotions of the youth generation. Strongly Positive items, numbering 122, were concentrated in milestone years of major national projects, epochal policies, and national commemorative events, reflecting the spreading of mainstream discourse into cyberspace. Strongly Negative items lowered to only 40, heavily attached to malignant public events, with a significant decline after 2010. Temporal evolution patterns indicated that from 2000 to 2009, net buzzwords were predominantly driven by negative events; from 2010 to 2018, mild entertaining and positive words rose across the span; from 2019 to 2026, a diversified and balanced pattern emerged, featuring mainstream positive narratives, youth self-mockery and catharsis, and specialized digital terminology side by side. This research revealed the deep interlocking logic behind the emotional shifts in internet language over 27 years, involving media iteration, social governance, youth mentality, and national narratives, which drove Chinese cyberculture to disseminate with the balanced development of the closed-loop transmission path of national hardware and software.

1. Introduction

With the deepening advancement of its reform and opening-up as well as science and technology, China has gradually realized the Internet of Everything and is moving toward the Intelligent Internet of Everything ¹. People use network terminals such as smartphones and computers to access the internet conveniently, simultaneously and efficiently, which has boosted the development of online language and facilitated communication both domestically and globally. As a special linguistic form born out of the online environment, internet buzzwords keep evolving alongside social events and internet technologies, and have now become a vital cultural symbol that reflects social mentality and collective emotions ². Throughout the popularization of the internet in China, internet buzzwords have grown from sporadic trendy terms in the early days into a complete discourse system, from coded slang for forum discussions in the BBS era into a cultural landscape covering all areas of society. There exists a covariational relationship between language and society, which means that language as a social phenomenon invariably changes along with gradual or drastic transformations in social life ^{3, 4, 5, 6}. In such a sense, internet buzzwords are not merely a passport to the online world, but also a cultural code that embodies the sentiments of the times ⁷. With a keen insight into society, they capture the shifts in the collective psychology of netizens.

Nowadays, emotion is regarded as an important analytical dimension for interpreting internet buzzwords. For one thing, buzzwords are inherently produced and adopted for an emotional relief to evoke resonance in emotion and will ⁸. For another, as a distinctive means of social interaction and netizen participation, internet buzzwords play various roles and represent personal feelings for identity recognition and emotional resonance ⁹. They thus serve as a vivid window into social mentality ¹⁰.

Different from the natural renewal of conventional words, the emotional connotations of internet buzzwords have undergone striking historical shifts including semantic generation mechanisms and communication dynamics of buzzwords. Nevertheless, the long-term researches on them as a continuous emotional sequence are urgent and necessary. There is also a lack of systematic exploration into the phased patterns of emotional evolution and the underlying driving forces behind such shifts. Accordingly, the theoretical and practical developments of internet buzzwords need to be clarified, which are related to the questions: what are the evolutionary trajectory and features of Chinese internet buzzwords in past decades? Driven by what social structural and mass media technological factors have the emotional expressions and shifts of internet buzzwords occurred? And/or what about profound changes in the socio-psychological structure behind them by constructing a diachronic corpus and integrating emotional analysis with discourse analysis?

2. Literature Review

Lingual variation is the inevitable result of language development during social progress. Language and society maintain a covariational relationship, and the social implications embedded in language can reflect social dynamics ^{11, 12}. Language originates and evolves within social communication activities, serving as a unique tool for large-scale interpersonal interaction ¹³. The systematic correspondence and interactive development between linguistic forms and social variables cause language variation to manifest in multiple dimensions ¹⁴. Generally speaking, language variation refers to the formal and functional differences of language arising from ethical, social, cultural, regional, psychological, contextual and other factors in a speech community from synchronical and diachronical development ^{15, 16, 17}.

People communication and language contact constitute the crucial external motivation triggering linguistic development and variation. When people move from one location to another, dialect contact and cross-cultural communication directly influence the magnitude of linguistic variation, which verifies the spatio-temporal attributes of migrant trajectories ¹⁸. Cross-cultural contact reveals the restructuring mechanism of linguistic structural systems driven by cultural integration and providing empirical evidence for language evolution amid globalization. Dialect contact deepens and expands linguistic scopes beyond traditional phonetics to lexical and grammatical domains, facilitating the systematization of dialectal variation. Moreover, specific social groups also possess the intrinsic correlations between linguistic variation and identification approval. Dialect speakers bear a prominent correlation between language usage and ethnic identity after dialectal contact ¹⁹. Distinctive variational features tend to be shaped by specialized terms including rigorous syntactic structures, unique cultural contexts, and occupational and/or age apply paradigm.

The Internet has extended the field of language variation from offline to online, giving rise to new lingual features. In contemporary online linguistic practice, internet neologism has become a typical sociolinguistic landscape. It carries ideological and emotional forces that co-vary with the social environment, reflects the prevailing social structure and relevant concrete knowledge, and fosters psychological and emotional identity among specific groups ²⁰.

In terms of variation types, internet language variation mainly falls into three categories: phonetic homophony substitution, mixed use of words and letters, and simplified grammatical structures, which result from the joint effect of digital communication efficiency and expressive innovation ^{21, 22}. Regarding dynamic evolution, the micro-language chain variation path illustrates the complete life cycle of new media language including initiation, variation, diffusion and multi-stage evolution, featuring circle-based communication and coordinated pragmatic ecology ²³. As for specific phenomena, there appears an evaluation system for the vitality of language variation based on the annual top ten internet buzzwords ²⁴.

The prevalence of a large number of internet buzzwords constitutes a history of contemporary social mentality recorded by linguistic forms, reflecting the collective emotions of special social groups under social pressure in specific historical periods. In this process, internet buzzwords have evolved from a mere communication tool into a means of constructing group boundaries. As a kind of special linguistic folklore, a collection of traditional customs and stories pass through the generations developing from by word of mouth in face-to-face communication to by type of hands with internet connection ²⁵. Net buzzwords have undergone four evolutionary stages: the rise, the boom, the transformation, and the continuous iteration, which clearly demonstrates the interaction between language variation and community identity.

Behind online words and variation with inherent instability lies the relation of power of different subjects. According to the perspective of Foucault’s theory, online linguistic variation demonstrates the developing game of linguistic power between main-stream dominant discourse and sub-cultural discourse. Online linguistic variation with asymmetric power relations and ideological influences emerges from discursive struggles among multiple power subjects ²⁶. Linguistic variation in online buzzwords is never an autonomous internal evolution of linguistic systems, but a meaning-making practice deeply embedded within social activities and structures. Meanwhile, linguistic variation coincides with the construction of community identity: netizens assimilate emerging ideas in cyberspace and gradually diverge from one another; driven by the innate need for internal belonging, they coalesce into distinct subcultural cohorts as subtle resistance against mainstream formal discourse. Once these buzzwords gain popularity, mainstream media and societal forces incorporate such subcultural terms by stripping their original subversive connotations and reshaping semantic meanings, triggering secondary semantic shifts. In turn, niche variants construct communal identities, and divergent group affiliations further fuel continual evolution of internet language, following a full-cycle logic of discipline-resistance-incorporation-identity construction.

The emotional attribute of language is not an inherent characteristics of words but rather gradually emerges through the interplay of social interaction, linguistic variation and group representation construction. Social representation theory was firstly proposed by French socio-psychologist Serge Moscovici in 1961 ²⁷. Its core definition is mainly defined as a gradually generated system of shared cognition, collective emotion, and consensual symbolic codes within a specific community through daily social interactions, media dissemination, and real-life experiences. It represents the psycho-discursive product through which a group transforms unfamiliar social phenomena into comprehensible common sense. Distinct from the individual psychological paradigm that focuses on private cognition, this theory moves beyond the scope of individual mental analysis, emphasizing that representations are not isolated subjective thoughts but public knowledge rooted in the social environment and formed through collective negotiation. They are externally materialized into linguistic symbols such as vocabulary, idioms, and internet catchphrases ^{28, 29}. From the perspectives of linguistics and new media studies, a group’s social representations are directly reflected in the evolution of linguistic semantics and affective connotations, making the theory a crucial analytical tool for explaining the emotional shifts and semantic variations in internet slang. Recent domestic research has further localized and refined the concept: in digital media environments, internet users rely on social platforms to co-construct representations, and online expressions have become the most intuitive material carriers of social representations ^{30, 31}.

The development of social representations theory can be divided into three periods: classical foundation, structural expansion, and localized application. The first period is its classical foundation from 1960s to 1990s when the theoretical prototype was proposed based on empirical research into the psychoanalysis. There are two core mechanisms which are anchoring and objectification, thereby breaking away from the individual-experiment bias that dominate psychology at that time and constructing an analytical chain of “society–group–cognition–symbol” ²⁹. Period Two is its structural paradigm development from 1990 to 2020, which is signifies as the core-periphery dual-structure model with a stable core and a flexible periphery ³². The core consists of long-lasting mainstream values of a speech community, determining the essential nature of the representation while the periphery adjusts flexibly in response to social currents and environmental changes, allowing semantic and emotional shifts. This model enhanced the theory’s dynamic explanatory power ³³. The last period presents the digital and localized development from 2020 to now. Duo to the booming of science and technology and the proliferation of social media, the theory has been applied to the study of internet language and youth subcultures all over the world ^{34, 35}.

The media ecology of communication offers a unique technological-cultural analytical framework for understanding the emotional expressions in online buzzwords. This proposition “the medium is the message” asserts that what truly matters is not the specific content carried by the medium, but the medium itself—its formal characteristics determine how information is perceived, understood, and disseminated ^{36, 37, 38, 39}. Applied to online language phenomena, this means that changes in digital emotional expression are not merely superficial updates in content, but rather a profound reshaping of social interaction patterns and emotional expression modes driven by the evolution of media technology.

Viewed through media ecology, a clear evolutionary trajectory emerges in the development of internet buzzwords as three stage model of toy, mirror, art, providing a powerful analytical framework for this analysis ^{40, 41, 42}. In the early stage of online language development, new media technologies were often regarded as toys, with internet language emerging in playful, fragmented forms. As these technologies evolved into mirrors, online language began to fully reflect societal contradictions and psychological states. Finally, in the artistic stage, internet language achieves a return to the fundamental roots of emotional expression. This three-phase progression reveals a novel pathway for linguistic and cultural dissemination: the construction of media environments in the toy phase, the formation of content frameworks in the mirror phase, and the reclamation of social emotional expression in the art phase. This process profoundly illustrates how technological evolution gradually reshapes the possibilities of emotional expression.

Specifically, the advancement of network technology has brought about fundamental transformations in emotional expression. From 1G to 5G, the rise of short-video platforms has made internet slang increasingly colloquial and visual, characterized by distinct fragmentation and emotional intensity. Nearly half of the annual internet buzzwords initially arise from short-video platforms. The shaping effect of technological media on emotional expression is evident not only in the shift toward visualized language forms but also in deep changes to the rhythm and intensity of emotional communication—moving away from traditional logical expressions toward fragmented, entertainment-driven linguistic inflation ⁴³. Notably, image symbols possess associative connotations whose meanings are often fluid, ambiguous, and uncertain, enabling them to escape, to some extent, the constraints of conventional text-based censorship mechanisms ⁴⁴. In this sense, media technological development does more than provide new tools for emotional expressions, it also transforms the status and function of emotional communication within the broader social interaction system. Internet language represents not merely an innovation in communication tools, but a transformation in ways of thinking and experiencing emotion in the native intelligence of cyberculture ⁴⁵.

Discourse analysis theory (DAT) provides a profound theoretical resource for interpreting emotional expressions and power operation of online buzzwords, which concludes discourse-power theory of Foucault and critical discourse analysis of Fairclough. Preciously, discourse analysis (DA) focuses on pure linguistic contexts, including semantic structures, conversations, narrative styles, genre of writing and so on ^{46, 47}. In the early years Foucault introduced a political dimension into the traditional linguistic domain famously emphasizing the power of discourse ^{48, 49}. It is regarded that discourse is not merely a linguistic concept but a third entity beyond mere speech and language—power generates discourse, and discourse in turn produces power. Foucault’s conception of power originates from institutions; it is a relational concept that cannot exist independently, but rather manifests through concrete mechanisms of control and counter-control ⁵⁰.

With Halliday’s Systemic Functional Linguistics (SFL) proposed, lingual forms become meaningful within the application of societal communication, achieving a transformation from structure to contextual function ^{51, 52}. Fairclough launched the study of discourse criticism, with the core goal being to reveal the class, race, and political inequality of discourse reproduction, forming three internationally recognized analytical paradigms ⁵³. At present, multimodal discourse analysis with quantitative and qualitative mixed researches has become the mainstream methodology ⁵⁴. Society as a whole is embedded within an all-encompassing network of power, with discourse serving as its external expression. From this theoretical perspective, internet buzzwords is no longer simply a record of technological evolution, but a significant indicator of shifting discursive power relations in cyberspace. Moreover, internet buzzwords not only reflect the dissolution of traditional discursive power and the construction of netizens own discursive authority, but also, with the emergence of numerous virtual communities, signal a trend shift in online discourse power from a differential structure toward a collective one. The production and application of online buzzwords carry inherent power dynamics, simultaneously disrupting traditional social symbolic systems and reflecting ongoing changes in contemporary social power relations and structures.

3. Research Methodology

In order to get enough data of online buzzwords available, the research was designed within the time span from 2000 to 2026 with official resources of mainstream media platforms, including People’s Daily, China Daily, CNKI, Yaowenjiazi, Bilibili, Weibo, Douyin, and Jinri Toutiao, and so on. One of Chinese academic authorities is China National Knowledge Infrastructure (CNKI), which provides access to information resources, knowledge dissemination and digitalization all over the world ⁵⁵. Every year these media platforms publish the annual buzzwords that recur frequently across them ²⁴. These are officially recognized with the highest weight. There are also some salient commercial platform popularity sources to obtain annual trending buzzwords from API or annual reports of Weibo, Bilibili’s annual live-chat highlights, and Douyin/Today’s Headline’s annual hot topics. The size was set at the top 20 most popular buzzwords each year.

A standardized corpus of emotional evolution in online buzzwords was constructed, including: basic information about each entry including name, year of emergence, origin, meaning, usage context; annual frequency of use, dissemination platforms, and distribution across subcultures; key features such as sentiment labeling, sentiment intensity, and nodes of sentiment shift.

The data processing was divided into online buzzwords collection, initial screening and standardization with cleaning. The first step was to choose the online buzzwords to categorize into annual sets from those platforms mentioned above for initial screening. The second step is to unify their format to remove topic tags like # or *, to convert all English letters to lowercase and all Chinese characters to simplified form. Chinese is a logographic language, in which there are no capital letters for application habits ⁵⁶. The third step was to lexical merging to consolidate different forms of a term under a single main entry, which removed the identical and similar ones. The forth step was structural integrity check to eliminate incomplete lexical fragments with the same meaning for detection is less effective for sentences with complex semantic information especially those with too few sentiment words ⁵⁷.

Then the most crucial step was to label the online buzzwords with multimodal emotional analysis from Valence (Positive/Negative/Neutral/Complex), Intensity with polarized class (1-5) and Emotion Types (Anxiety, Banter, Self-mockery, Identification, Anger, Comfort, etc.) ⁵⁸.

From the internet platforms mentioned above, there were about 20 words collected in each year from 2000 to 2026, so that there were 540 online buzzwords in total in the list as the database.

To mark out the different degree of each online buzzwords from emotional analysis. The core reference sources were the classification from HowNet Emotional Dictionary ^{59, 60} and Dalian University of Technology Information Retrieval Laboratory (DUTIR), and Chinese Sentiment Lexicon ^{61, 62}, combined with emotional intensity gradient rules, adapted through a five-level hierarchical approach.

To address characteristics of internet buzzwords, such as irony, self-deprecation, meme culture, social sentiment, and event-derived terms, two additional evaluation dimensions are introduced as semantic context and public emotional tone. About annotation guidelines, following the principle of semantics-first with usage context as supplementary, judgments were made by integrating word’s literal meaning, online extended meanings, public emotional connotations, and dissemination atmosphere, clearly distinguishing between literal definitions and actual emotional expressions. To grade the five tiers, emotions were categorized into 1-5 levels with the sequence of online buzzwords, presence and intensity and nature of emotion like irony, dissatisfaction or anger, and so on. Higher numerical grades indicated stronger emotional intensity, greater conflict and more extremity.

There are principles of initial screening needed to follow. The first principle is about synonymy. Some online buzzwords are put into the lexical entry, for they are connected by slashes with similar meanings or semantic structure. Like: 玉米/凉粉/笔迷, which means the big fans of some celebrities. Additionally, core terms are extracted from long phrases, and redundant modifiers are removed. The second principle is cross-year deduplication, which means for identical buzzwords with exactly the same meaning across different years, only the first occurrence is retained, and all subsequent duplicates are deleted; variations in meaning or usage evolution are not duplicated. The third one is to integrate the buzzwords with similar wording and highly overlapping meanings, which need to be merged into a core term, avoiding duplicate counting. The last one is to distinguish the buzzwords with derivative structures, which belong to semantic extensions, so that the main term is preserved, and derivative terms are merged under it.

After the initial screening of online buzzwords from 2000 to 2026, there were 13 terms of completely repeated cross-year vocabulary, 2 terms of repeated internal-year terms, and 3 groups of highly overlapping and homologous semantic words were cleaned. In total, 18 redundant terms were eliminated out from the initial database. Therefore, there are 522 buzzwords left in total as the refined data to further analysis.

According to the selected foundational emotional lexicon database, HowNet Emotional Lexicon is used to distinguish between positive, negative, and neutral buzzwords with intensity scores attached; DUTIR subdivides them into 7 major emotional categories (joy, fondness, anger, sadness, fear, disgust, surprise), and accompanied by intensity gradients (levels 1-5); Chinese Emotional Lexicon is for the supplement with dedicated sentiment annotations for internet colloquialisms and internet memes.

Five-Tier Hierarchical Emotional Valence System was utilized as the following table:

There is adaptation of special principles for online buzzwords. The first is template patterns (for example: “XX Gate”, “XX on the Tip of the Tongue, etc.), which is divided into the set of Neutral (N3) for the template itself carries no emotion; emotion is determined by the content that fills the slot. The second is self-deprecating buzzwords (for example: involution (内卷)、lying flat (躺平)、 a big foll (大冤种),etc.), which are uniformly classified as Weakly Negative (N2). Moreover, terms related to contemporary policies, sports event slogans, and major national achievements are set as Strongly Positive (P5) while those entries concerning disasters, social conflicts, and right-defending tragedies are regarded as Strongly Negative (N1); buzzwords associated with entertainment, food, light-hearted memes, and aesthetics belong to Weakly Positive (P4); professional economic/technical jargon, special names of online platforms, and objective event nouns are categorized into Neutral (N3).

Based upon Table 3, the total number of final entries is 806 (122 + 247 + 221 + 176 + 40 = 806). The five emotional labels demonstrate unique characteristics as following:

Strong Negative (N1) is entirely connected with malignant social incidents, public disasters, and illegal phenomena: “SARS”, “a mining accident”, “hide-and-seek” (a prison abuse case), “gutter oil”, “bullying and hegemony”, “deep fakes”, and so on, which exhibit strong event-dependency. Before 2010, the volume of strong negative online buzzwords was significantly higher than that in later years; with social governance improved, derived buzzwords from major incidents decreased substantially.

Weak Negative (N2) serves as the primary carrier for netizens to release their emotions: from 2007-2010, the complaints about rising prices; from 2019-2022, youth survival anxieties, for example: “involution”, “lying flat”, “mental exhaustion”, “wage earner”, “corporate slave”, and so on. These are mainly self-mocking and stress-relieving expressions, lacking intense confrontation, and representing a kind of emotional compensation among younger demographics.

Neutral buzzwords (N3) visibly fall into three categories. The first category is basic internet terminology, like BBS, QQ, blogs, metaverse, AI-related words; the second is solid template phrases, like “A Bite of XX”, “ineffective XX” and so on; the last category is the objective words from economy, industry, and academia. For example: “Three Rural Issues”, “low-altitude economy”, “MBTI”, etc. From 2024 to 2026, the share of neutral terms continues to rise, with a substantial increase in objective terminology related to AI and the digital economy.

Weak Positive (P4) with the number of 247 maintains the highest proportion over the long term, spanning the entire period: early-stage net celebrities and leisure hobbies; mid-stage short-video memes and aesthetic appreciation; later-stage healing content, leisure travel, and cute/cuddly terms, like “Hajimi”, “meal partner”, “dopamine dressing”, and so on. These reflect the mainstream emotional needs of netizens in daily entertainment and aesthetic expression.

Internet buzzwords from the last column of Strong Positive (P5) surge with waves of bigger occasions during major milestone years: 2005 (Olympic slogan), 2008 (Beijing Olympics), 2012 (the Chinese Dream), 2017 (the New Era), 2021 (the centenary of the founding of the CPC), and 2023 (New Quality Productive Force), and so on. Core sources include national mega-projects, sporting events, governance philosophies, the main themes of the era, and terms evoking national pride, which are typical manifestations of official discourse spreading all over the internet space.

Furthermore, there is temporal evolution trends of net buzzwords. From 2000 to 2009, negative emotional buzzwords stood out, with dense clusters of strong negative terms; net buzzwords were largely triggered by social negative events. On the other side, positive buzzwords centered on major public events such as space missions and the Olympics in Beijing. From 2010 to 2018, comparably weak positive entertainment-oriented buzzwords exploded; net celebrities, comedy sketches, and short-video memes became mainstream all over China. Oppositely, self-deprecating weak negative terms grew slowly, and strong negative terms declined obviously. From 2019 to the present, a clear binary divergence emerges: on the one hand, positive buzzwords as the mainstream continue to be steadily produced, while on the other hand youth-anxiety-related weak negative buzzwords surge. Simultaneously, AI and the digital industry introduce a large number of neutral professional buzzwords, causing the overall emotional structure more diverse and balanced.

4. Conclusion

After conducting multi-layer data cleaning, online buzzwords from 2000 to 2026 cover four major eras of Chinese Internet development from online public discourse, social hot topics, policy narratives, youth subculture and industrial technology, which guarantees sufficient long-term sequential representativeness. Based on HowNet Emotional Dictionary, DUT Sentiment Vocabulary Ontology Database and Chinese Sentiment Lexicon, a five-tier emotional valence framework is adopted to classify all buzzwords into five categories: Strongly Positive (P5), Weakly Positive (P4), Neutral (N3), Weakly Negative (N2) and Strongly Negative (N1). The static distribution and long-term evolutionary trends of each category are summarized as follows:

Firstly, Weakly Positive buzzwords (247 entries) account for the largest proportion and serve as the mainstream daily emotional carrier of netizens. This category includes recreational memes, aesthetic compliments, leisure and healing online expressions covering the whole 27 years. It demonstrates that the core demand of residents using the Internet lies in entertainment, aesthetic expression and mild positive emotional release, and lightweight optimistic sentiment forms the fundamental emotional tone of online language.

Secondly, Neutral buzzwords (221 entries) keep expanding, especially after 2024. Neutral words contain three sub-types: basic internet media terminology, fixed template sentence structures and objective nouns of economy and technology. The booming of AI, low-altitude economy and digital industry generates a large number of neutral professional vocabularies, which proves that digital technology and industrial discourses have continuously spread online public space, enhancing the objective and professional attributes of internet slang.

Thirdly, Weakly Negative buzzwords (176 entries) act as the primary emotional compensation channel for young groups, mainly consisting of self-deprecation, helplessness and mild complaints without intense confrontation. Price-related complaints prevailed from 2007 to 2010, while anxiety-driven terms such as involution, lying flat and mental internal consumption surged during 2019-2022, which reflects young people’s mild way to relieve realistic pressure via sub-cultural self-mockery.

Fourthly, Strongly Positive buzzwords (122 entries) are closely tied to milestone national events, carrying national identity and mainstream official discourses. Concentrated in years with major national projects, sports events and policy releases, these words reflect the integration of official mainstream narratives and grassroots online expression, conveying national pride and recognition of the times.

Fifthly, Strongly Negative buzzwords (40 entries) are exclusively derived from public disasters, social conflicts and illegal incidents with strong event dependency. Such terms appeared densely before 2010; with the improvement of social governance, hot words stemming from vicious public incidents decreased sharply, and extreme negative sentiments faded away in online public discussions.

From the sequential evolution perspective, the emotional characteristics of net buzzwords can be divided into three distinct stages. During 2000-2009, negative sentiments dominated, and most buzzwords originated from adverse social incidents, with positive vocabularies limited to a small number of national major events. In 2010-2018, entertainment-oriented Weakly Positive expressions became mainstream, extreme negative vocabularies declined significantly, and mild self-deprecating negative words grew slowly, shifting online language from critical discussion to daily entertainment. From 2019 to 2026, an obvious binary divergence emerges together with expanding neutral technical terms: mainstream Strongly Positive narratives are steadily produced, while youth anxiety-related Weakly Negative buzzwords remains prevalent; massive neutral digital technology words enrich the whole database, forming a diversified and balanced emotional structure of online language.

In general, the emotional evolution of Chinese online buzzwords from 2000 to 2026 is a comprehensive reflection of internet media iteration, social governance progress, youth mentality changes and the popularization of national development narratives. Online emotional expression has transformed from sharp, event-driven negative catharsis in the early Internet era to lightweight recreational positive creation, and finally developed into a balanced system with stable mainstream positive discourse, mild youth emotional releasing channels and supplementary objective industrial terminology. Extreme negative sentiments gradually recede, and mild, diverse and daily-oriented expression becomes the core emotional feature of contemporary internet buzzwords.

This research reveals the deep interlocking logic behind the emotional shifts in internet language over 27 years, involving media iteration, social governance, youth mentality, and national narratives, which drives the buzzwords to disseminate with the balanced development of the closed-loop transmission path of national hardware and software, which leads to another trend that national confidence is promoting and more attention is paid towards international affairs.

5. Limitations

Though this research was under conduction with concerns and plannings, there inevitably existed some limitations as following:

Firstly, this research selected approximately 20 representative online buzzwords each year to construct a 540-item corpus, which was not enough and need capture the complete set of annual buzzwords across the entire network in the future. There was a subjective preference in the screening process. Some niche memes and sub-cultural vocabulary specific to certain circles were not included in the research. The corpus was biased towards popular media-released mainstream buzzwords, lacking coverage of the emotional expressions of netizens from specific circles. It was unable to fully reflect the emotional differences within the network of different circles.

Secondly, the research ended in the first four months of 2026 with buzzwords, which leaded to the result of 2026 net buzzwords without a complete annual propagation cycle verification. And some buzzwords had a narrower dissemination scope and lower public recognition than the mature buzzwords from previous years, and the stability of emotional annotation was weaker than the those from 2000 to 2025.

Thirdly, the collected online buzzwords lacked regional and platform stratification comparisons, and it was difficult to reveal the differential influence of media fields and regional cultures on buzzword emotions.

Last but not least, internet buzzwords tended to acquire polysemous connotations across practically diverse contextual scenarios, accompanied by dynamic shifts in their emotional valence, which rendered the original emotional annotation framework rigid and lacking in contextual adaptability.

Funding

This study was supported by Research Project of Henan Provincial Social Science Association under Grant No. SKL-2025-1489.

References