A Study on the Impact of New Media on College Students’ Oral English Learning
—Taking English Content Creators on Bilibili as an Example ()
1. Introduction
In the era of global digitalization, the digital video industry has experienced rapid growth. According to the 52nd Statistical Report on China’s Internet Development, as of June 2025, China’s internet user base reached 1.123 billion, with internet penetration rising to 79.7%. The number of online video users stood at 1.044 billion, representing a usage rate of 96.8% among internet users, demonstrating robust growth momentum. Among these, Bilibili, a short-video platform originating from anime and manga culture, has gained immense popularity among the youth of Generation Z. According to its Q2 2024 financial report, the platform recorded 477 million monthly active users (MAU) during the quarter, a 10% year-over-year increase, and 141 million daily active users (DAU), up 11% year-over-year. These Tables underscore Bilibili’s leading position in the short video industry, providing a solid real-world context for this analysis.
With the continuous evolution of technology, traditional learning methods are being progressively disrupted by digital innovations. Students are no longer confined to conventional learning models; an increasing number are accessing real-time educational content through short-video platforms to enhance learning efficiency. Bilibili, a short-video platform characterized by interactivity, diversity, and user-generated content (UGC), offers English learners a wide array of learning opportunities. Currently, the author conducted combined searches using the keywords “Bilibili” and “English speaking skills” in databases such as CNKI and Wanfang. The most frequently appearing keywords were “Bilibili” (465 instances) and “Bilibili bullet comments” (143 instances). However, relevant literature on English speaking skills within new media, the focus of this study, was scarce (58 instances). Furthermore, existing research predominantly focuses on describing learning behaviors, lacking a standardized evaluation system for oral proficiency improvement. This deficiency undermines the comparability and persuasiveness of research conclusions. Given this, the author adopts internationally recognized English speaking proficiency standards as the core analytical framework. Five key dimensions—phonetics and intonation, fluency and coherence, lexical resource, language behavior competence, and discourse appropriateness—along with specific observation points are defined. This establishes a unified basis for precisely evaluating the impact of Bilibili creators on speaking learning in subsequent analysis, while also pointing out concrete directions for enhancing college students’ speaking abilities. This approach enhances the research’s targeted and operational nature.
2. Overview of Bilibili in Language Teaching
Bilibili’s short videos initially focused on anime, merchandise, movies, and live streaming, gradually expanding into fields like education, technology, and knowledge. Today, it has evolved into the primary online gathering place for Gen Z youth. With the surge of Gen Z users on Bilibili and mounting pressures in modern society, short video consumption has shifted from leisure entertainment to self-improvement and spiritual enrichment. Consequently, short videos have become one of the platforms supporting Gen Z youth in their learning journey.
2.1. Characteristics of Language Teaching on Bilibili
Language serves as a bridge connecting the world, underpinning the foundations of human society and profoundly shaping cultural, cognitive, social, and individual development. At every level, language forms the foundation of human civilization and social progress. Yet it remains only a vague reflection within human thought; only through systematic language learning can the emotions and intentions people cannot voice aloud be woven together. As China’s second-largest short-video platform, Bilibili offers diverse and innovative language teaching content that provides effective pathways to enhance students’ linguistic abilities and spark learning enthusiasm. This is made possible by its vibrant user base and robust community of creators. Such development responds to contemporary demands, delivering richer, more profound learning experiences. Compared to traditional English teaching models, Bilibili’s approach fosters greater student autonomy, convenience, and learning interest. The author selected several influential English learning creators from Bilibili, whose language teaching approaches encompass the following characteristics.
1) Diverse Creative Content:
From the creators’ perspective, different content creators exhibit varied teaching styles.
First: TED Super Speaker creators feature uniqueness, relatability, focus, emotional appeal, and humor in their teaching. They offer three viewing modes—bilingual, English-only, and subtitle-free—to meet learners’ personalized needs.
Second: YouTube Spoken English Selection creators emphasize creativity, presenting diverse English series across various topics. This provides learners with effective channels for acquiring knowledge while broadening their existing knowledge base.
Third: Lin Xiaolu English Intensive Reading focuses on serving students preparing for postgraduate entrance exams, CET-4/6, TOEFL, and IELTS. Through intensive reading of foreign publications, it breaks down grammar challenges, tackles complex sentences, and helps grasp article structure and logic, comprehensively enhancing learners’ language proficiency.
Fourth: Mmxgxg creators feature authentic video learning, emphasizing not only spoken English but also cultivating English thinking patterns to avoid Chinese-influenced expressions.
In summary, students can select their preferred English learning approach from creators with distinct styles based on personal preferences. This personalized learning path helps enhance students’ interest in learning.
2) Diversification of Teaching Models:
In the era of big data, traditional teaching methods struggle to meet the demands of the new landscape, necessitating the integration of conventional approaches with new media. bullet comments serve not only as a medium for expressing personal opinions but also as a pathway for students to acquire knowledge. They enable learners to address questions encountered during video-based learning through real-time interaction, thereby enhancing educational efficiency. This model aligns with the concept of “social scaffolding” in Vygotsky’s theory of psychological development. Knowledge is transmitted from more capable individuals to less capable ones through bullet, creating a supportive structure akin to scaffolding in construction [1]. This support extends beyond knowledge transfer to include emotional support; Bilibili’s real-time interaction feature enables learners to engage with content creators or fellow learners while watching videos. Through bullet comment interactions, students can instantly access needed knowledge. This community engagement not only strengthens connections among learners but also provides timely feedback and Q&A opportunities, thereby enhancing learning outcomes.
3) Diverse Student Choices:
The multi-dimensional search capabilities of online learning platforms generally meet the retrieval needs of students with varying learning objectives. For instance, content can be filtered by duration, view count, release date, and other criteria. After entering keywords, students can further refine their search to access a wealth of relevant learning materials. Regarding resource quality, Bilibili offers an extensive and diverse array of English learning resources, covering nearly all domains and difficulty levels of English study, providing students with a broad spectrum of choices.
2.2. Influence of Bilibili Oral English English Content Creators
Bilibili’s low barrier to entry has attracted a large influx of content creators, forming a diverse and vibrant creative community. This has also fostered a multitude of outstanding creators across various fields. For instance, in the language learning domain, the author has identified the following four influential spoken language creators through a series of rigorous criteria, including subscriber count, official certification, total views, total likes, membership level, and number of subscribers.
The data in Table 1 demonstrates that these four content creators have attracted
Table 1. Influence data of Bilibili English content creators.
Content Creator |
YouTube Spoken English Highlights |
TED Super Speaker |
Lin Xiaolu English Intensive Reading |
Mmxgxg |
Subscribers (10k) |
542 |
432.3 |
270.4 |
142.7 |
Total Views (10k) |
9872.5 |
3425 |
6752 |
4476 |
Total Likes (10k) |
285.5 |
111.8 |
212.2 |
117.4 |
Officially Verified |
Yes |
Yes |
Yes |
Yes |
Member Level |
6 |
6 |
6 |
6 |
Charging Supporters (persons) |
273 |
107 |
269 |
194 |
widespread attention by providing high-quality learning resources through their unique styles, distinctive content, and deep linguistic expertise.
3. Current Issues in Modern College Oral English Teaching
In the new era, accelerated economic and social development coupled with increasingly frequent international exchanges have led to a growing demand for foreign language talent across all industries. However, among the four skills of listening, speaking, reading, and writing, contemporary university students’ oral proficiency still falls short of meeting the nation’s requirements for foreign language professionals. Professor Shao Yongzhen, after conducting a survey in 1998 to revise the university English teaching syllabus, concluded that: only 5% of university graduates in recent years possessed strong or very strong oral skills, while 37% had poor or very poor skills. Only 7% were competent or basically competent to participate in international conference discussions, and only 14% were competent or basically competent to engage in foreign business negotiations [2]. When searching for “CET-4/6 oral” on Xiaohongshu, the most common comments were: “What’s the point of CET-4/6 oral?”, “Is it worth signing up for the oral?”, and “Can I skip the exam after registering?” These questions indirectly reveal Chinese students’ notably low priority on oral skills. Since most universities link graduation to CET-4/6 written scores, this creates a phenomenon of high scores but low practical ability. The issues above primarily involve the following aspects.
1) Regarding University Students
Oral communication issues primarily manifest in two aspects: lack of confidence and difficulty in communication. First, due to the influence of traditional modest cultural values and learning models, Chinese students generally exhibit characteristics of self-deprecation, deference, and a strong desire to save face [3]. Consequently, they often lack the courage to speak up in English classes. Second, individual differences and proactive engagement significantly impact oral progress. China’s current English education model prioritizes exam-oriented teaching, viewing test scores as the sole measure of proficiency. Finally, Chinese English instruction heavily emphasizes input while neglecting output, with insufficient post-class practice. This results in inadequate oral skills, making “silent English” a stark reality in China’s spoken English learning landscape.
2) Teacher Perspective
Traditional English speaking instruction often prioritizes test-oriented teaching, neglecting the communicative essence of spoken language. Teachers equate oral instruction with pronunciation correction and sentence pattern memorization, overemphasizing grammatical accuracy and collocation while neglecting the communicative function and real-world application of spoken language. Furthermore, this teaching model suppresses student agency. Classroom time is predominantly allocated to teacher demonstrations and knowledge transmission, with student speaking time falling below 30%. Interaction is largely confined to question-and-answer formats.
3) Educational Resources
The uneven distribution of teaching resources is one reason for the current challenges in oral language instruction. First, some schools have relatively scarce resources for oral language teaching, lacking support from real-life contexts, which limits students’ ability to apply their skills in daily communication. Second, the outdated nature of textbooks and multimedia resources fails to reflect the contemporary characteristics of language, making it difficult for students to keep pace with new developments in modern language [4]. Finally, outdated educational philosophies that overemphasize exam performance while neglecting practical application have significantly contributed to a severe polarization in language proficiency, substantially undermining the cultivation of college students’ oral communication skills.
4. Research Design and Methods
To verify the impact of Bilibili English content creators on college students’ spoken English learning and whether they can enhance students’ linguistic and pragmatic abilities, this chapter will detail the overall research design framework and specific implementation strategies. Through scientific methods, the proposed practical issues will be transformed into measurable empirical analyses, thereby ensuring the reliability and validity of subsequent data analysis. This chapter will clarify the core research questions, the specific hypotheses derived from these questions, the composition of the survey participants, and the research methods and tools employed, laying the foundation for the empirical analysis throughout the paper.
4.1. Research Questions
Addressing the challenges in college students’ spoken English learning—such as lack of contextual exposure, insufficient motivation, and inadequate resource suitability this study systematically examines the impact of Bilibili English content creators as a new media platform. The core research questions are: 1) Is there a positive correlation between Bilibili UP owners’ content and the spoken English improvement of their subscribing users (college students)? 2) Can college students enhance their oral language competencies (phonetics/intonation, fluency, vocabulary) and pragmatic abilities (language behavior skills, discourse appropriateness) through new media?
4.2. Research Hypotheses
Bilibili English content creators exhibit a positive correlation with college students’ English speaking proficiency improvement. This relationship is validated through three dimensions: speaking ability enhancement, learning motivation, and learning satisfaction improvement, as well as content consumption behavior. Hypothesis 1: A positive association between content consumption behavior and speaking ability improvement (linguistic competence and pragmatic competence). Hypothesis 2: A positive association between content consumption behavior and learning motivation and satisfaction.
4.3. Research Subjects
The survey participants were undergraduate students enrolled at Shaanxi University of Science and Technology, covering disciplines such as science and engineering, humanities and history, and arts to ensure the representativeness and coverage of the sample. In terms of gender distribution, 180 male respondents (60%) and 120 female respondents (40%) participated. Regarding academic year, sophomores and juniors collectively accounted for 61%, while freshmen, seniors, and graduate students comprised the remaining 39%.
4.4. Research Methods
To scientifically validate the impact of Bilibili English content creators on college students’ spoken English learning and clarify their role in enhancing linguistic and pragmatic abilities, this study adopts a mixed-methods research approach combining questionnaire surveys, data analysis, and descriptive analysis. The overall framework centers on internationally recognized spoken English proficiency standards (see Table 2 for specific criteria). This framework provides the sole standardized basis for designing questionnaire items, data coding, and subsequent result analysis within the self-perceived oral proficiency dimension, ensuring consistent and systematic evaluation of “oral proficiency” throughout the research process. This guarantees both the systematic nature of the research process and the reliability of the findings.
The questionnaire design revolves around the core research questions and hypotheses, encompassing four interrelated dimensions: First, the personal information dimension collects foundational data such as respondents’ gender, grade level, and English proficiency, laying the groundwork for subsequent sample characteristic analysis and result interpretation. Second, the content usage behavior dimension of UP owners covers four sub-dimensions—usage frequency,
Table 2. Oral English proficiency evaluation standards and observation dimensions.
Competence Categories |
Core Dimensions |
Specific Observation Points |
Linguistic Competence |
Pronunciation and Intonation |
Accuracy of phoneme pronunciation; appropriateness of word stress and sentence intonation; liaison and weakening in speech flow. |
|
Fluency and Coherence |
Stability of speech rate and rhythm; frequency of hesitation and repetition; use of logical connectors and discourse markers. |
|
Vocabulary Resources |
Richness and accuracy of vocabulary; authentic use of idioms, slang, and collocations; ability to avoid repetitive wording. |
Pragmatic Competence |
Linguistic Action Ability |
Ability to realize speech functions such as requests, suggestions, apologies, and objections in specific contexts. |
|
Discourse Appropriateness |
Awareness and ability to choose formal/informal styles based on social distance, power relations, and communication occasions; sensitivity to cultural norms. |
content preferences, interaction depth, and learning methods—to quantify the specific behavioral characteristics of college students engaging with Bilibili’s English learning content. Third, the self-perceived oral proficiency dimension includes four core observation points: pronunciation accuracy, expression fluency, vocabulary flexibility, and cross-cultural communication awareness. This aims to compare changes in students’ abilities before and after exposure to content creators. The “pre-exposure” data was collected through retrospective self-assessment by respondents. Considering the practical challenges of longitudinal tracking in large-scale surveys, this study employs retrospective evaluations based on respondents’ memory of their pre-exposure oral proficiency. It should be objectively noted that such retrospective self-reported data may suffer from validity limitations including memory bias, diminished recall accuracy, subjective cognitive interference, and inconsistent evaluation standards over time. However, subsequent reliability and validity tests have verified the overall stability and effectiveness of the questionnaire, mitigating these limitations to a certain extent. Fourth, the learning motivation and satisfaction dimension measures the impact of UP creators’ content on college students’ intrinsic motivation for oral learning and user feedback.
During the data processing phase, this study employed SPSS 26 statistical software for quantitative analysis: a paired-sample t-test was conducted to compare differences in self-perceived oral proficiency before and after exposure to UP core content, thereby testing Hypothesis H1; correlation analysis was used to clarify the strength of associations among usage behavior, perceived oral proficiency, learning motivation, and satisfaction. Multivariate linear regression analysis was employed to investigate the predictive role of each sub-dimension of usage behavior on spoken language proficiency improvement, quantifying core influencing factors. This approach comprehensively addressed the research questions and validated Hypothesis H2, establishing a clear technical pathway for subsequent detailed data analysis and result interpretation.
5. Data Analysis and Results
A total of 330 questionnaires were distributed, with 300 valid responses collected. Data analysis was conducted using SPSS 26 statistical software. Through reliability and validity testing, the scientific rigor of the questionnaire was verified. Subsequently, core data analysis, correlation analysis, and regression analysis were conducted across three dimensions: “Bilibili English Content Creator Usage Behavior”, “Self-Perceived English Speaking Ability”, and “English Speaking Learning Motivation and Satisfaction”. This process systematically validated the influence of Bilibili English content creators on college students’ English speaking learning, providing empirical support for the research hypotheses.
5.1. Descriptive Statistical Analysis
Descriptive statistical analysis revealed 300 valid questionnaires, comprising 180 males (60%) and 120 females (40%). By grade distribution: 88 were sophomores (29.33%), 95 were juniors (31.67%), 43 were seniors (14.33%), and 22 were graduate students or above (7.34%). The majority were sophomores and juniors, aligning with the core demographic for college students learning spoken English. Regarding English proficiency, 156 students (52%) had passed CET-4, 78 (26%) had passed CET-6, 32 (0.67%) had not passed CET-4, and 34 (11.33%) were taking specialized English or IELTS/TOEFL exams. The sample exhibited a tiered distribution of English proficiency levels.
5.2. Reliability and Validity Testing
To ensure the reliability of research conclusions, the author conducted rigorous quality testing on the measurement tools employed. The results of reliability and validity analysis indicate that this questionnaire can stably and effectively measure the actual impact of Bilibili content creators’ content on college students’ spoken English learning. This lays a methodological foundation for subsequent in-depth analysis of its positive effects on learning behaviors and attitudes.
5.2.1. Reliability Testing
Reliability testing aims to evaluate the stability and consistency of measurement results. The author employed Cronbach’s alpha coefficient for this assessment. Analysis revealed an overall reliability coefficient of 0.923 for the questionnaire. Reliability performance across dimensions was as follows: the alpha coefficient for the “Bilibili English Content Creator Usage Behavior” dimension was 0.886, for the “Self-Perceived English Speaking Ability” dimension was 0.901, and for the “English Speaking Learning Motivation and Satisfaction” dimension was 0.897. All coefficients significantly exceeded the benchmark of 0.70, indicating the questionnaire’s internal consistency and the reliability of the collected 300 sample data points. This result confirms, from a measurement methodology perspective, the credibility of students’ feedback regarding their BiliBili-based spoken English learning behaviors, perceived ability changes, and motivational experiences. It provides a robust data foundation for subsequent research revealing the positive learning effects facilitated by new media platforms.
5.2.2. Validity Testing
Validity testing focuses on whether the questionnaire accurately measures the core concepts the researcher intends to explore. The study first assessed data suitability for factor analysis using the KMO test and Bartlett’s sphericity test. The results were satisfactory: the KMO value was 0.865, and Bartlett’s sphericity test yielded a chi-square value of 2863.521, significant at the 0.001 level. This fully demonstrates the suitability of the sample data for factor analysis. Following exploratory factor analysis on each of the questionnaire’s three dimensions, all items under each dimension clearly aligned with their respective single common factors. Factor loadings ranged from 0.723 to 0.891, indicating a clear structural configuration. This outcome robustly confirms the questionnaire’s validity. The design aligns with the research framework, accurately capturing and measuring core variables such as college students’ behavioral characteristics in learning spoken English through Bilibili creators, their perceived skill improvement, and shifts in affective attitudes.
Combining the reliability and validity test results, reliability ensures the stability of measurement outcomes, while validity demonstrates that the questionnaire effectively measures the impact of new media usage on spoken English learning outcomes. This indicates that the empirical data obtained through this questionnaire authentically and effectively reflects the multifaceted positive effects of Bilibili creators on college students’ spoken English learning. Subsequently, based on this high-quality data, the hypotheses proposed in this study will be thoroughly validated.
5.3. Core Data Analysis Across Dimensions
This study employs mean (M) and standard deviation (SD) as descriptive statistical indicators to characterize the distribution patterns of each core dimension (using a 5-point Likert scale: 1 = Disagree, 2 = Somewhat disagree, 3 = Neutral, 4 = Somewhat agree, 5 = Agree).
5.3.1. Content Usage Behavior Dimension of English Content Creators
The “Bilibili English Content Creator Usage Behavior” dimension encompasses four sub-dimensions: usage frequency (X1), content preference (X2), interaction depth (X3), and learning approach (X4). The overall mean for this dimension is 3.56 (standard deviation = 0.78, standard error = 0.045), falling between “Moderate” and “Fairly Consistent”. This indicates that college students exhibit a clear positive tendency toward using Bilibili English content creators’ materials, with their behavior demonstrating a degree of regularity and proactivity. In terms of data scientificity, the absolute values of skewness and kurtosis for all items and overall dimensions were less than 0.5, conforming to an approximate normal distribution. The standard deviations of each item ranged from 0.76 to 0.95, with standard errors between 0.044 and 0.055. The data exhibited low dispersion and high stability, objectively reflecting the general characteristics of the group (see Table 3).
Data shows Q5 mean 3.89, the highest across all dimensions, indicating over half of college students have established a fixed learning frequency, incorporating it into routine study habits; Q6 mean 3.72 reflects targeted usage centered on goals like “improving pronunciation” and “building vocabulary”; Q7 and Q8 means of 3.68 and 3.65, respectively, exceed the overall dimension average, highlighting students’ emphasis on learning continuity and content specialization. Q9 (mean 3.59) and Q10 (mean 3.47) scored at moderate levels. While some students have moved beyond “one-way viewing” by interacting to resolve doubts or bookmarking for review, the means remain below 3.6. This indicates insufficient adoption of interactive and review practices, with some students failing to establish consistent habits due to concerns about expression or lack of time. Q11 Mean
Table 3. Usage behavior analysis.
Item No. |
Mean (M) |
Standard Deviation (SD) |
Standard Error (SE) |
Skewness |
Kurtosis |
Q5 I regularly watch English-language content creators on Bilibili for spoken English practice, such as situational dialogues and pronunciation tutorials. |
3.89 |
0.76 |
0.044 |
−0.321 |
−0.156 |
Q6 I will select specific types of Bilibili English content creators based on my personal needs (such as improving pronunciation or building vocabulary). |
3.72 |
0.81 |
0.047 |
−0.289 |
−0.213 |
Q7 On average, I spend over 20 minutes watching spoken English content from Bilibili creators each time. |
3.68 |
0.79 |
0.046 |
−0.305 |
−0.189 |
Q8 I prefer to follow Bilibili English content creators with professional backgrounds in English education or extensive overseas experience. |
3.65 |
0.83 |
0.048 |
−0.278 |
−0.231 |
Q9 I will communicate with Bilibili English content creators about spoken English learning through comments, bullet comments, or private messages. |
3.59 |
0.85 |
0.049 |
−0.292 |
−0.201 |
Q10I save high-quality spoken English content from Bilibili creators for easy review and repeated learning. |
3.47 |
0.88 |
0.051 |
−0.267 |
−0.198 |
Q11 I will practice speaking by imitating the pronunciation, intonation, and logical flow of English content creators on Bilibili. |
3.32 |
0.91 |
0.053 |
−0.254 |
−0.224 |
Q12 I will develop a customized speaking practice plan incorporating the learning methods provided by the content creator (such as shadowing practice). |
3.15 |
0.95 |
0.055 |
−0.241 |
−0.217 |
Overall Dimension |
3.56 |
0.78 |
0.045 |
−0.296 |
−0.195 |
Note: Skewness absolute value < 0.5 and kurtosis absolute value < 0.5 indicate the data approximates a normal distribution. SE = SD/√n (n = 300).
3.32; Q12 Mean only 3.15, the lowest across all dimensions. This indicates that university students still primarily engage in “passive information reception”, lacking sufficient awareness to actively convert viewed content into practical exercises. There is a deficiency in the deep transformation process from input to output.
Overall, college students’ usage behavior exhibits proactive input and relatively passive output: On one hand, they can systematically and purposefully acquire high-quality resources based on their own needs, demonstrating the autonomy and flexibility of new media learning; On the other hand, they exhibit significant shortcomings in deep engagement (interaction) and skill transfer (imitation, planning), failing to fully leverage Bilibili’s strengths in community interaction and personalized learning. Their usage remains confined to “superficial consumption”, lacking the complete learning cycle of reception, interaction, practice, and reflection. Future efforts should focus on strengthening proactive imitation, deep engagement, and personalized planning to elevate their experience from passive viewing to active learning.
5.3.2. Self-Perceived Oral Proficiency
Based on the English oral proficiency evaluation framework, this section examines changes in students’ abilities before and after exposure to UP content across two domains: linguistic competence and pragmatic competence. A paired t-test was employed to analyze these differences and test Hypothesis H1. The specific results are presented in Table 4:
Table 4. Self-efficacy analysis.
Ability Dimension |
Pre-Exposure M ± SD |
Post-Exposure M ± SD |
Mean Difference (MD) |
t-value |
P-value |
Cohen’s d (Effect Size) |
Pronunciation Accuracy (Stress, Linking) |
2.41 ± 0.83 |
3.85 ± 0.75 |
1.44 |
23.67 |
<0.001 |
1.72 |
Expression Fluency (Frequencyof Hesitation) |
2.35 ± 0.85 |
3.78 ± 0.78 |
1.43 |
22.91 |
<0.001 |
1.67 |
Vocabulary Flexibility |
2.52 ± 0.81 |
3.82 ± 0.76 |
1.30 |
21.34 |
<0.001 |
1.56 |
Cross-Cultural Communication Awareness |
2.28 ± 0.87 |
3.69 ± 0.80 |
1.41 |
20.78 |
<0.001 |
1.51 |
Overall Dimension |
2.43 ± 0.84 |
3.76 ± 0.77 |
1.33 |
25.12 |
<0.001 |
1.83 |
Data indicates that after exposure to Bilibili English content creators, college students achieved statistically significant improvements across all core dimensions of spoken English proficiency (all p-values < 0.001). The most pronounced improvement occurred in “pronunciation accuracy” (mean difference = 1.44, Cohen’s d = 1.72). This aligns with the design logic in Table 2, which positions “phonetics and intonation” as a foundational language competency dimension. It confirms the reinforcing effect of Bilibili creators’ “pronunciation instruction” and “situational imitation” content on basic language skills. The improvement in the “Cross-Cultural Communication Awareness” dimension was equally significant (mean difference = 1.41), echoing Table 2’s emphasis on “Discourse Appropriateness” and highlighting the value of multicultural scenarios presented in UP owners’ content for cultivating pragmatic skills; Meanwhile, improvements in “fluency of expression” (mean difference = 1.43) and “flexibility of vocabulary use” (mean difference = 1.30) respectively align with the observational requirements for the “fluency and coherence” and “vocabulary resources” dimensions in Table 2. This demonstrates that UP-main content comprehensively covers the core components of oral proficiency, achieving synergistic development of linguistic competence and pragmatic skills. Overall, the average oral proficiency score post-exposure to UP-main content reached 3.76, representing a 1.33-point increase from the pre-exposure level (2.43). This clearly demonstrates the significant enhancement effect of Bilibili English UP-main content on oral proficiency, validating its practical value in improving college students’ foundational oral communication skills.
5.3.3. Learning Motivation and Satisfaction
This dimension aims to validate research hypothesis H2. The overall mean score for the “English speaking learning motivation and satisfaction” dimension was 3.76, approaching the “somewhat agree” level. This indicates that Bilibili English content creators’ materials exert a significant positive driving effect on college students’ English speaking learning motivation. Furthermore, students’ satisfaction with such content remains at a relatively high level. Statistical data is presented in Table 5.
Table 5. Learning motivation and satisfaction analysis.
Item Content |
Mean (M) |
Standard Deviation (SD) |
Standard Error (SE) |
95%Confidence Interval (CI) |
UP hosts’ content stimulates interest in speaking practice |
3.92 |
0.73 |
0.042 |
[3.84, 4.00] |
Willing to invest more time for better results |
3.75 |
0.78 |
0.045 |
[3.66, 3.84] |
Satisfied with UP host’s content quality |
3.81 |
0.75 |
0.043 |
[3.73, 3.89] |
Would recommend high-quality UP hosts to others |
3.68 |
0.80 |
0.046 |
[3.59, 3.77] |
Overall Dimension |
3.72 |
0.76 |
0.044 |
[3.63, 3.81] |
Analysis reveals that the overall mean score for the “English Speaking Motivation and Satisfaction” dimension is 3.72 approaching Fairly Agree, indicating that content creators’ material significantly stimulates college students’ speaking motivation and yields high user satisfaction. Among the items, content creators stimulates interest in spoken English learning had the highest mean score (3.92), significantly higher than the “passive learning” state before exposure. This indicates that the engaging and scenario-based nature of content creators effectively addresses the pain point of traditional spoken English learning being “dull and tedious”. The average score for “Satisfaction with Content Quality” reached 3.81, reflecting university students’ recognition of the creators’ content for its professionalism and practicality, further validating its positive value in spoken language learning.
5.4. Correlation and Regression Analysis among Variables
5.4.1. Correlation Analysis
To clarify the degree of association between “Bilibili English Content Creator Usage Behavior” and “Self-Perceived English Speaking Ability” as well as “Learning Motivation and Satisfaction”, pearson correlation coefficients were employed for analysis. The results are presented in Table 6:
Table 6. Three variables correlation analysis.
Variables |
Usage Behavior |
Oral English Self-Efficacy |
Learning Motivation and Satisfaction |
Usage Behavior |
1 |
- |
- |
Oral English Self-Efficacy |
0.683 |
1 |
- |
Learning Motivation and Satisfaction |
0.721 |
0.705 |
1 |
Note: indicates p < 0.01, indicating extremely significant correlation.
Data shows that the correlation coefficient between “Bilibili English content creator usage behavior” and “self-perceived English speaking ability” is 0.683 (p < 0.01), while the correlation coefficient with “learning motivation and satisfaction” is 0.721 (p < 0.01). Both exhibit moderately strong positive correlations, with extremely significant statistical significance. Simultaneously, “self-perceived speaking ability” and “learning motivation and satisfaction” also showed a significant positive correlation (r = 0.705, p < 0.01). This indicates that the more actively college students engage with content creators’ materials, the more pronounced their self-perceived improvement in speaking ability becomes, accompanied by higher learning motivation and satisfaction. These three factors form a positive feedback loop, further validating the positive impact of content creators’ materials.
5.4.2. Regression Analysis
To clarify the predictive role of “Bilibili English content creators’ usage behavior” on “self-perceived English speaking ability”, a multiple linear regression model was constructed. “Self-perceived speaking ability” served as the dependent variable (Y), while the usage behavior dimensions—“usage frequency (X1)”, “content preference (X2)”, “interaction depth (X3)”, and “learning method (X4)”—were used as independent variables. The regression analysis results are in Table 7.
Table 7. Three variables regression analysis.
Independent Variables |
Regression Coefficient β |
Standard Error |
t-value |
p-value |
VIF Value |
Usage Frequency (X1) |
0.285 |
0.062 |
4.59 |
<0.001 |
1.32 |
Content Preference (X2) |
0.312 |
0.058 |
5.38 |
<0.001 |
1.28 |
Interaction Depth (X3) |
0.198 |
0.065 |
3.05 |
0.003 |
1.35 |
Learning Methods (X4) |
0.256 |
0.061 |
4.20 |
<0.001 |
1.15 |
Constant Term |
0.523 |
0.189 |
2.77 |
0.006 |
- |
R2 |
0.482 |
- |
- |
- |
- |
F-value |
56.32 |
- |
- |
<0.001 |
- |
Note: All VIF values are < 2, indicating no multicollinearity issues.
The regression model results show that the overall F-value is 56.32 (p < 0.001), indicating good model fit. R2 = 0.482, meaning the four independent variables—“usage frequency”, “content preference”, “interaction depth”, and “learning method”—collectively explain 48.2% of the variance in “self-perceived oral proficiency”, demonstrating strong predictive power. Regarding individual variables, all regression coefficients were positive with p-values < 0.01 (p = 0.003 for interaction depth < 0.01). “Content preference” exhibited the highest regression coefficient (β = 0.312), followed by “Usage Frequency” (β = 0.285) and “Learning Approach” (β = 0.256). This indicates that college students’ proactive learning behaviors—selecting content based on personal needs, increasing usage frequency, and adopting imitation practice—most significantly drive oral proficiency improvement. These findings quantitatively validate the positive impact of Bilibili English content creators on college students’ spoken English learning.
Crucially, this study relies on cross-sectional data. The predictive effects identified through regression analysis represent statistical associations at a single point in time, not strict causality. Key limitations include:
1) Ambiguous causal direction: The model assumes “usage behavior leads to skill improvement”, but reverse causation may exist—students with stronger oral foundations and higher confidence may be more inclined to actively use UP creator resources. Existing data cannot distinguish between these confounding factors. 2) Lack of temporal validation: Oral proficiency improvement requires long-term accumulation. This study lacks longitudinal tracking, making it impossible to confirm whether usage precedes skill enhancement or to differentiate between short-term interest and long-term outcomes. 3) Omission of key variables: The model excludes confounding factors such as baseline English proficiency, self-discipline in learning, and the quality of university oral courses. These factors may simultaneously influence both usage behavior and skill improvement, leading to overestimation of regression coefficients. 4) Sample selection bias: Respondents were exclusively “active users of UP content”, potentially introducing a bias toward “intrinsically motivated learners”. 5) Common method variance: Both independent and dependent variables were self-reported, allowing students to subjectively link “high-frequency usage” with “skill improvement”, thereby amplifying statistical correlations. 6) Correlation does not imply causation: Regression reveals only the strength of association and cannot address intervention questions such as whether forcing increased usage frequency would enhance abilities. This differs fundamentally from the causal effects demonstrated in randomized controlled trials.
Mitigation strategies and recommendations for regression analysis limitations are as follows:
1) Employ longitudinal tracking designs: Collect data at multiple time points to validate temporal sequence between variables and clarify causal directionality; 2) Refine models and samples: Incorporate control variables like baseline levels and learning traits, include “non-user” control groups to balance baseline characteristics; 3) Optimize measurement methods: Use objective metrics such as standardized oral tests and Bilibili backend usage data to reduce subjective bias; 4) Define association boundaries: Characterize “predictive effects” as statistical correlations, explore mechanisms through qualitative research, and avoid causal inferences.
In summary, predictive effects provide quantitative references for associations between variables. However, constrained by cross-sectional data and confounding variables, they cannot be equated with causal conclusions. Future studies require more rigorous designs to validate the actual impact of usage behavior.
6. Conclusions and Recommendations
Through theoretical exploration and empirical analysis, this study has verified the influence mechanism of Bilibili English content creators on college students’ spoken English learning. This chapter aims to summarize the research findings presented throughout the paper, provide clear answers to the research questions posed at the outset, elaborate on the theoretical and practical significance of the study, reflect on its limitations, and point the way forward for future research in related fields. It also proposes corresponding learning strategies to enhance college students’ spoken English proficiency.
6.1. Research Findings
Based on a questionnaire survey and data analysis of 300 university students, the core findings of this study can be summarized in three aspects, fully validating the proposed research hypotheses:
First, Bilibili English content creators can effectively and comprehensively enhance university students’ spoken English proficiency, strongly supporting Research Hypothesis H1. Data analysis reveals statistically significant improvements across four core dimensions: vocabulary resources (flexibility in word usage), and the discourse appropriateness (cross-cultural communication awareness) of pragmatic abilities. The average improvement exceeded 1.3 points, with effect sizes (Cohen’s d) all greater than 1.5, indicating a large effect level. This fully demonstrates that Bilibili creators’ content precisely covers the key dimensions of spoken English proficiency. This addresses the first core question of this paper: Bilibili creators’ content exhibits a significant positive correlation with college students’ oral proficiency improvement. Furthermore, the findings indicate that through this new media channel, college students can not only enhance traditional language skills but also strengthen pragmatic abilities—crucial for authentic communication—thus achieving the coordinated development of comprehensive language literacy [5].
Second, Bilibili English content creators significantly boost college students’ learning motivation and enhance their learning satisfaction, providing empirical validation for research hypothesis H2. The study found that the overall mean score for the “English speaking motivation and satisfaction” dimension was 3.72, approaching the level of “somewhat agree”. Among the items, “UP owners’ content stimulates interest in speaking practice” scored the highest (M = 3.92), indicating that the engaging and scenario-based nature of their content effectively alleviates the tedium of traditional speaking practice, successfully transforming external learning behaviors into intrinsic motivation. Simultaneously, students demonstrated high satisfaction with content quality (M = 3.81). This finding reveals that Bilibili creators’ positive influence extends beyond cognitive skill enhancement, profoundly impacting learners’ autonomy.
Third, regarding usage behavior, skill enhancement and motivation satisfaction, and self-perceived competence—the three dimensions—correlation analysis indicates that each pair of these core variables exhibits a moderately strong positive correlation (r values ranging from 0.683 to 0.721). More in-depth regression analysis further revealed that among the sub-dimensions of usage behavior, selecting content based on personal needs (content preference, β = 0.312), maintaining regular usage (usage frequency, β = 0.285), and adopting active learning strategies such as imitation and shadowing (learning methods, β = 0.256) were the most critical factors predicting perceived improvement in spoken language proficiency. This indicates that strategic usage behaviors directly foster skill growth, while the resulting sense of accomplishment and the inherent appeal of content creators’ material jointly enhance learning motivation and satisfaction. These three elements mutually reinforce each other, forming a dynamic model of positive reinforcement.
It should be objectively noted that this study is subject to the limitation of recall bias. All data regarding participants’ oral proficiency and learning status prior to exposure to Bilibili’s English content were derived from retrospective self-assessments. Participants may have underestimated their past language proficiency due to perceived improvement in their current abilities, potentially leading to an overestimation of the effect size for self-perceived skill enhancement. However, this bias may only affect the magnitude assessment of the correlation effect, without altering the core finding that exposure to Bilibili English content creators positively correlates with college students’ spoken English proficiency and learning motivation. Future research could adopt longitudinal tracking designs, incorporating objective pre-baseline testing to mitigate recall bias and further validate these conclusions [6].
In summary, new media serves as a significant force in enhancing college students’ oral proficiency. Nevertheless, research conclusions should be interpreted rationally, accounting for limitations such as recall bias. Through more rigorous design and practical strategies, the pedagogical value of content creators can be further explored to advance oral teaching reform.
6.2. Recommendations
To further leverage the positive impact of Bilibili English content creators on college students’ spoken English learning, targeted optimization suggestions are proposed from both student and creator perspectives.
1) Student Level
Students should develop personalized learning plans based on their oral communication weaknesses, aligning with the core dimensions outlined in Table 2: For the “Pronunciation and Intonation” dimension, prioritize content creators specializing in pronunciation instruction and shadowing practice to reinforce foundational skills like phoneme articulation and stress patterns through repeated exercises; For weaknesses in “Discourse Appropriateness”, select creators covering diverse scenarios like business communication, daily conversations, and cross-cultural exchanges. Scenario-based learning enhances register selection and cultural sensitivity. For weaknesses in “vocabulary resources” and “fluency and coherence”, leverage creators’ thematic teaching content to accumulate authentic vocabulary combinations [7]. Simultaneously, improve expression fluency through shadowing and spontaneous retelling to achieve balanced enhancement across all competency dimensions.
2) Creator Level
Creators can optimize content creation based on the evaluation framework in Table 2:
For linguistic competence: Continuously strengthen foundational teaching in pronunciation, intonation, and vocabulary usage while designing targeted practice scenarios.
For pragmatic competence: Further enrich cross-cultural communication scenarios and refine explanations of discourse appropriateness to help learners translate language skills into practical communication abilities, thereby enhancing teaching effectiveness.