Patterns and Explanatory Factors of Language Proficiency in a Sample of Secondary School Students, with Special Regard to the English Language

Abstract

The factors determining the patterns and levels of individuals’ foreign language skills form a complex system that collectively shapes their effectiveness. However, the use of the target language today takes place in a completely different environment and form, partly due to digitalization. Our study focuses on the patterns of language proficiency among secondary school students (spoken languages, types of spoken languages, self-reported level of English proficiency, and number of English language exams) and everyday forms of target language use. The research sample consists of two waves of a national Hungarian survey (Hungarian Youth Research 2016, Hungarian Youth 2020). In the case of the 2020 database, we ran two regression models to explain the level of at least intermediate knowledge of English and the existence of English language exams. The results show the dominance of English and its consolidation between 2016 and 2020, while the number of spoken languages has decreased slightly. At the same time, there has been a positive change in the level of language proficiency. At least intermediate proficiency in English is embedded in a wide range of explanatory variables. The factors influencing the possession of language exam certificates differ from the explanatory variables of language proficiency.

1. Introduction

Language proficiency is a key factor in labor market success (John et al., 2021; Mavisakalyan, 2017; Beblavy et al., 2016), but its level is embedded in several background factors (DeKeyser, 2013; Duff, 2019; Darvin & Norton, 2016) and is not solely dependent on individual performance and skills. The novelty of our study is that it examines the variables explaining the number of spoken languages in a regression model (sociodemographic background variables, type of secondary school, ways of using the target language, geographical indicators, etc.) and compares the background factors of language exams and self-reported language proficiency using a Hungarian research database. Our analysis focuses on the sociology of education, which is also reflected in the background variables used. In our work, we used the Hungarian Youth Research 2016 and Hungarian Youth 2020 databases, from which we separated the secondary school age group. Using these two databases, we were able to conduct a longitudinal study of elements of language proficiency such as spoken languages, the number of language exams taken, self-reported language proficiency, and changes in the use of the target language. The relevance of our study is not only due to the complexity of the model used but also to the fact that language proficiency in Hungary is characterized by strong inequalities and an unfavorable pattern when compared to other European countries.

2. Literature Review

2.1. The Sociological Context of Language Proficiency

Success in language learning and the results achieved by language learners in this field can be influenced by a number of factors. One group of influencing factors can be described from a sociological perspective (Norton & Toohey, 2011), while other approaches emphasize psychological foundations (Dörnyei & Ryan, 2015). One of the most researched areas of sociology is the reproduction of social layers and social mobility. It is fundamental that different groups reproduce to a significant extent (Breen & Goldthorpe, 1997), and language skills play an important role in this reproduction, either as a skill required for certain jobs or as a requirement for admission to higher education and for passing language exams that are requirements of obtaining a degree. The ownership of foreign language books, parents’ language skills, a positive attitude towards school and learning, and the use of extra language lessons are all related to the level of foreign language proficiency and the effectiveness of language learning.

Social background variables are related to students’ foreign language performance in several ways. Parents’ educational background is an outstanding factor, as research and review studies conducted in different countries consistently suggest that children of more highly educated parents generally achieve better foreign language results (Abbasian et al., 2020; Butler & Le, 2018; Szabó et al., 2021). The family’s socioeconomic status (SES) is also a determining factor in students’ foreign language development. Disadvantaged learners often have less access to learning resources, quality teaching materials, and supplementary language learning opportunities, which has been shown to be associated with poorer language learning outcomes (Nieścioruk, 2024; Zou et al., 2007). Low SES is also associated with lower levels of basic skills necessary for language development, such as working memory and grammatical sensitivity, which further hinders foreign language acquisition (Szabó et al., 2019). Several empirical studies have shown that students from less affluent families underperform in receptive English language skills, particularly in reading and listening comprehension (Abbasian et al., 2020; Zou et al., 2007). Cultural capital—primarily reading habits at home—also contributes to vocabulary and reading comprehension development (Dong & Chow, 2022; Lau & Richards, 2021). Demographic factors—primarily gender and age—also play a role in language learning, as there are differences in motivation, communication behavior, and certain skills (Baker & MacIntyre, 2000; MacIntyre et al., 2003; Kheder & Rouabhia, 2023) and in the age-specific characteristics of pragmatic norm acquisition (Derakhshan et al., 2024). Although social background plays a significant role, there are both high-achieving and low-achieving learners in every social group (Zou et al., 2007), which is why targeted support and a supportive pedagogical environment are of paramount importance in mitigating disadvantages (Ma et al., 2022; Ariani & Ghafournia, 2016). Overall, we can say that language learning is, among other things, a socially embedded process in which individual characteristics and environmental factors both determine learning outcomes (Norton & Toohey, 2011; Dörnyei & Ryan, 2015).

2.2. Extramural Activities and Language Proficiency

The rapid spread of digital technology has created a new type of linguistic socialization environment for learners: in their everyday lives, they are constantly exposed to multimodal English-language content, which significantly expands their language learning opportunities outside of school (Godwin-Jones, 2018; Soyoof et al., 2023; Maristy, 2023; Inayati et al., 2025). This trend is also supported by student experiences during the pandemic, which report increased online exposure to English (Canrinus et al., 2024). Research on mobile technologies also shows that these devices create new learning contexts and provide flexible, continuous access to foreign language content that can effectively support language skill development (Viberg & Grönlund, 2012). Previous research shows that extramural and informal foreign language input—especially regular reading and listening to audiobooks, as well as consuming foreign language media content—support the development of vocabulary and receptive language skills (Webb & Chang, 2015; Peters & Webb, 2018). Interactive, meaning-making activities such as video games or online communication, which can be considered more productive forms of extramural English language use, can also be linked to the development of productive language skills, particularly speaking skills (Sundqvist, 2009). These activities often strengthen learners’ motivation and autonomy, as they take place in interest-driven, authentic, and informal contexts (Soyoof et al., 2023). Based on the latest research trends, extramural English use is increasingly taking place in a digital form and is organically linked to learners’ everyday language practices. At the same time, several areas—such as the analysis of different age groups or the possibilities for pedagogical integration—remain partially unexplored (Inayati et al., 2025; Maristy, 2023). Differences in social background and individual characteristics continue to play an important role in language performance differences, but research suggests that these factors interact with the quantity and quality of extramural language use, which appears to be a decisive variable in itself (Cadierno et al., 2022; Paradis, 2011).

2.3. European and Hungarian Context

The European Council considers foreign language communication to be one of the key competences that is essential in our globalized world in order to find a “common voice” not only with the inhabitants of the countries of the European Union but also with other nations. The European Union, of which Hungary has been a member since 2004, clearly advocates multilingualism because, in its view, this serves the interests of the European economy and promotes communication and understanding between nations (Council of the European Union, 2019). The action plan of the European Commission was therefore drawn up with the aim of extending the benefits of language learning as a lifelong activity to all European citizens and ensuring a favorable environment for language teaching. However, looking at European countries, we can state that despite long-term plans, there are still a few countries whose inhabitants are leaders in this respect and who, by their own admission, speak at least two foreign languages in addition to their mother tongue (European Commission, 2012). In the European Union, the Baltic states (Latvia, Lithuania, and Estonia) reported speaking at least two foreign languages in the self-reported survey (52.6%, 47.7%, and 43.2%), while among Hungarian respondents, this proportion was only 14.2% in 2022. However, compared to the 2016 statistical results, it can be concluded that the language skills of Hungarians aged 25–64 have improved by 3.1% in recent years (Eurostat, 2024b).

Data on the language skills of Hungarians show that even our knowledge of the first foreign language is below the European average (54%). According to Eurobarometer data from 2012, 35% of respondents said they could speak a foreign language at a conversational level (European Commission, 2012). According to a survey conducted by the Statistical Office of the European Union in 2016, this ratio has changed slightly. Statistical data at that time showed that 42% of Hungarians aged 25–64 spoke a foreign language, but Hungary still ranks among the lowest in this respect, with only the United Kingdom and Romania behind us (Eurostat, 2018). Eurostat’s (2022) data also show similar ratios: three years ago, 41.1% of Hungarians aged 18–24 said they spoke a foreign language (Eurostat, 2024a, 2024b). It is important to note that the country is essentially monolingual, with no significant minority groups speaking other languages. At the same time, due to the selective education system, the obtaining of language exam certificates plays an important role in young people’s further education. The number of languages taught in secondary school is determined by the type of school (regular high school or vocational high school). Advanced language classes (that started after the change of regime in 1989) or special classes offering CLIL programs are typically attended by young people from favorable backgrounds (Fekete & Csépes, 2018; Szabó et al., 2021). Overall, we can say that language skills, the number of languages spoken, and the obtaining of language exam certificates are strongly linked to inequalities.

Einhorn’s (2012) study examines the effectiveness of Hungarian language teaching in an international comparison, highlighting the developments that have taken place since the change of regime and the existing challenges and perceived problems. The increase in the number of lessons and the emergence of intensive language programs can be considered significant steps forward, but the measures have not always produced the desired results. One of the main problems is that due to the exam-oriented nature of the system, most language learners see the acquisition of a language exam certificate as the ultimate goal of language learning rather than viewing the target language as a means of personal and professional development and communication. Due to the lack of active practice of the foreign languages they have learned, many students have only passive knowledge.

As in many other countries, in Hungarian public education, the success of foreign language learning is determined not only by individual factors but also by school and social factors (Vajnai et al., 2022; Sebestyén, 2023). The dominance of English is clear, with most research focusing on the learning and teaching of this foreign language, while other languages, such as German or French, are less prominent, which may distort the picture of language learning efficiency (Józsa & Nikolov, 2005). The relationship between school type and foreign language performance is confirmed by empirical research (Vígh, 2013; Novák & Fónai, 2020). Vígh’s (2013) qualitative study based on intermediate-level English and German school-leaving exam results showed a significant difference in the performances of grammar school and vocational grammar school students. Sebestyén (2017) emphasizes education policy decisions in relation to German language school-leaving exam results (the process of admission to higher education, pre-scheduled school-leaving exams). In their questionnaire-based research, Novák and Fónai (2020) identified the same correlation between self-assessment and the possession of a language exam certificate. Social status also plays a key role in language acquisition, with students from higher socio-cultural backgrounds typically having greater access to quality language education and extracurricular language learning opportunities (Bors et al., 1999; Sebestyén & Hegedűs, 2017). Parents’ educational attainment and regional location are also correlated with language performance: students living in more developed areas achieve better results, while foreign language learning is less successful in disadvantaged regions (Sebestyén & Hegedűs, 2017). The educational level stays in close relationship with the number of spoken languages in Hungary in the adult population too (Central Statistical Office, 2022).

2.4. Research Questions

In our analysis, we formulated the following research questions:

R1. What differences can be observed between the two waves of the survey (2016; 2020) in terms of language proficiency patterns (spoken languages, number of spoken languages, and number of language exam certificates)?

R2. What factors explain at least intermediate English language proficiency in the 2020 survey?

R3. What factors explain the attainment of at least an elementary English language exam certificate in 2020?

3. Materials and Methods

3.1. Dataset and Used Variables

In our analysis, we used the Hungarian Youth Research 2016 and Hungarian Youth 2020 databases, from which we extracted subsamples of full-time secondary school students (N = 1619; N = 1587). Sampling was multistage and stratified, with the first stage consisting of a stratified sample of Hungarian settlements and the secondary frame consisting of young people living in those settlements. The sample was weighted by gender, age group, region, settlement type, and educational attainment (Domokos et al., 2021). In our study, we used the following variables to assess language proficiency in both databases: the type of languages spoken (in our analysis, we examine the percentage of respondents who speak English, German, French, Spanish, Italian, Romanian, and Russian at least at an elementary level), the number of languages spoken, the self-reported level of proficiency, and the existence of language exams (at least at an elementary level). The 2020 questionnaire assessed the use of foreign languages in four categories: learning, reading books and articles in print or online, watching films or series, and keeping in touch with friends. Respondents could choose from the categories never, rarely, often, and always. The 2016 survey only included the response option of studying, so we were only able to compare this one variable longitudinally. We used SPSS 28 statistical software for this analysis. The research was conducted by Társadalomkutató Ltd., 3 Ida Street, Ground Floor 1, 1143 Budapest, Hungary.

3.2. The Process of the Analysis and Sample Characteristic

In the first step of our analysis, we compared the percentage distributions of the aforementioned variables using the two databases to examine the shifts that can be observed in the area of language proficiency. In the next step, we ran two regression models on the 2020 database, with the dependent variables being at least intermediate English language proficiency and the existence of an English language exam. The choice of English was justified by the results of our research and by the fact that English dominates in everyday language use (reading the news and watching TV series) (Sundqvist, 2009; Leppänen et al., 2009). The independent variables were as follows: gender (dichotomous variable, 0 = female and 1 = male), three categories of subjective financial situation (negative, average, and good, with dummy coding, where the reference category was the negative pole), mother’s and father’s degrees (dichotomous variable, 0 = none and 1 = yes), type of settlement (village, small town, and county seat/capital, with dummy coding, where the reference was capital city), economic situation of the region (0 = more favorable and 1 = less favorable; the two categories are based on Eurostat (2022) regional purchasing power parity, where regions below 60% of the European Union average were classified as disadvantaged), and engagement in extramural English language use (learning, reading, watching series and films, and keeping in touch) as a dichotomous variable, where 1 indicates often or always and 0 indicates never or rarely. The sociodemographic variables cover the basic dimensions of social stratification (gender, subjective financial situation, father’s and mother’s educational level, type of settlement, and the economic situation of the region), and with the help of extramural activities (learning, reading, watching films or TV series, and keeping contact in a foreign language), the way of using a foreign language can be identified. With this method, the effects of sociodemographic and extramural factors can be measured at the same time in the case of at least intermediate English language proficiency and an English language exam. The distributions of the variables used in the regression models are shown in

Table 1. Sample characteristics based on variables used in regression models (Hungarian Youth 2020).

	Participants	Percentage
Sociodemographic variables
Gender
Male	848	53.4%
Female	739	46.6%
Subjective financial situation
Good	1081	70.7%
Average	387	25.3%
Disadvantaged	61	4%
Father’s educational level
Degree	1223	80.5%
No degree	297	19.6%
Mother’s educational level
Degree	1245	80.4%
No degree	304	19.6%
Settlement type	310	19.5%
Budapest (capital)	331	20.9%
County seat	522	32.9%
Small town	423	26.7%
Village
Economic situation of the region	814	53.1%
Not disadvantaged	773	48.7%
Disadvantaged	310	19.5%
Extramural activities
Learns in a foreign language
Does not/never/rarely speaks a foreign language	794	50.1%
Often/always speaks a foreign language	792	49.9%
Reads in a foreign language
Does not/never/rarely speaks a foreign language	1236	77.9%
Often/always speaks a foreign language	351	22.1%
Watches films or TV series in a foreign language
Does not/never/rarely speaks a foreign language	1188	74.9%
Often/always speaks a foreign language	399	25.1%
Keeping contact in a foreign language
Does not/never/rarely speaks a foreign language	1372	86.4%
Often/always speaks a foreign language	215	13.6%

. The mean of age is 16.75, and the standard deviation is 1.53.

Initially, we also wanted to include the type of school in the models (vocational training without a high school diploma or vocational training with a high school diploma). However, the very low number of language exams taken in vocational training without a high school diploma (only one in the research sample) made this idea impossible. Therefore, we decided not to examine the effect of school type in this study.

4. Results

As a first step in our analysis, we examined which languages high school students speak and what changes can be observed between the two waves of the study. In both waves, English was spoken by the majority, followed by German; however, the four-year period showed the growing dominance of English. Knowledge of the other languages included in our study was negligible (English: 72.6% in 2016 and 74.9% in 2020, German: 35.0% and 28.2%, French: 3.3% and 3.3%, Italian: 2.4% and 2.0%, Spanish: 1.9% and 1.3%, Russian: 0.7% and 0.3%, and Romanian: 0.3% and 0.3%).

Due to the low number of respondents, we examined the subjective language proficiency level of English and German speakers (with response options of elementary, intermediate, advanced, and native speaker level). The results are shown in

Table 2. Subjective assessment of language proficiency in English and German in 2016 (N = 1176 and 567) and 2020 (N = 1118 and 447), showing percentage distributions.

	English		German
	2016	2020	2016	2020
elementary	57.9%	54.4%	69.7%	63.4%
intermediate	39.5%	41.5%	29.2%	34.5%
advanced	2.4%	3.9%	0.8%	2.1%
native speaker	0.2%	0.2%	0.3%	0.0%

. It is apparent that between the two waves of the survey, the intermediate and advanced categories expanded for both languages, while the proportion of elementary level speakers decreased.

We also examined the proportion of language exam certificates in relation to the proportion of speakers of the given language. The number of speakers of a given language varies considerably, so we present the data for languages other than English and German for informational purposes only (

Table 3. Percentage of speakers of a given language who have passed at least an elementary language exam (N = 1176 and 1118).

	2016	2020
English	18.8%	21.8%
German	9.9%	17.1%
French	11.9%	18.9%
Italian	4.6%	10.2%
Spanish	6.0%	10.6%
Russian	7.8%	0.0%

). The rate of conversion of language skills into language exam certificates increased between the two waves.

In our analysis, we also examined the number of languages spoken. In 2016, 15.1% of high school students said they did not speak any languages, 52.9% spoke one, 30.1% spoke two, 1.8% spoke three, and 0.1% spoke four languages. The data for 2020 show a slight change, as students are more likely to know one language (13.6% speak no languages, 61.8% speak one, 23.7% speak two, 0.7% speak three, and 0.2% speak four). Taking the average number of languages spoken, there was a slight decrease between the two waves (2016: 1.18 and 2020: 1.12, standard deviations 0.66 and 0.63).

We were only able to compare language use in the case of “learning”, as the wording of the other categories differed between the two surveys. In this case, we found that in 2020, the values for the “rarely” and “often” categories were higher, while the proportions at the extremes decreased (2016—never: 16%, rarely: 23%, often: 39.1%, and always: 22%; 2020—never: 13.9%, rarely: 28.1%, often: 41.5%, and always: 16.2%). In the 2020 database, we were also able to examine other activities related to extramural English language use. The data obtained are presented in

Table 4. Other language-related activities in the 2020 sample (N = 1371), given as percentages.

	Reading Books or Articles Online or on Paper	Watching Films or TV Series	Keeping Contact with Friends
never	41.6%	39%	53.8%
rarely	32.5%	31.6%	30.1%
often	20.9%	23.6%	13.2%
always	4.7%	5.5%	2.5%

The table does not include the categories “don’t know” and “no answer”.

. The highest frequencies are associated with “never” responses: 42% never read in English, 40% never watch movies or TV series, and 54% never communicate in a foreign language.

The dependent variable in our first regression model was at least intermediate English language proficiency, which we measured using self-reporting.

Table 5. Regression model of background factors for at least intermediate English language proficiency (p < 0.05, significant correlations are marked in bold).

	B	S.E.	Wald	df	Sig.	Exp(B)
Sociodemographic variables
Gender (0 = female; 1 = male)	−0.150	0.124	1.480	1	0.224	0.860
Subjective financial situation (reference: negative financial situation)
Above-average financial situation	0.870	0.305	8.121	1	0.004	2.386
Average financial situation	0.276	0.162	2.907	1	0.88	1.318
Father has degree (0 = no; 1 = yes)	0.406	0.185	0.837	1	0.028	1.501
Mother has degree (0 = no; 1 = yes)	0.789	0.185	18.195	1	<0.001	2.201
Age (continuous variable)	0.226	0.040	32.190	1	<0.001	1.254
Type of settlement (dummy, capital city being the reference category)
County seat	0.454	0.213	4.517	1	0.034	1.574
Small town	0.190	0.191	0.990	1	0.320	1.209
Village	0.152	0.205	0.549	1	0.459	1.164
Economic situation of the region (0 = favorable; 1 = disadvantaged)	−0.266	0.142	3.508	1	0.061	0.766
Extramural activities
Learns in a foreign language (0 = no; 1 = yes)	0.384	0.137	7.835	1	0.005	1.468
Reads in a foreign language (0 = no; 1 = yes)	0.909	0.179	25.798	1	<0.001	2.482
Watches films in a foreign language (0 = no; 1 = yes)	0.858	0.174	24.421	1	<0.001	2.359
Keeps contact in a foreign language (0 = no; 1 = yes)	−0.216	0.205	1.100	1	0.294	0.806
Constant	−0.099	0.766	63.379	1	<0.001	0.002

2 Log likelihood: 1616.516; Cox and Snell r square: 0.193; Nagelkerke r square: 0.267.

shows the results of the statistics run on the 2020 database. The significant correlations cover a wide range of background variables (sociodemographic and extramural variables at the same time: above average financial situation, both parents’ education level, age, county seat, learns in a foreign language, reads in a foreign language, and watches films in a foreign language).

In the final step, we ran an explanatory model for the possession of at least an elementary level of English language proficiency) (

Table 6. Regression model explaining factors influencing English language exam results (p < 0.05, significant correlations are marked in bold).

	B	S.E.	Wald	df	Sig.	Exp(B)
Sociodemographic variables
Gender (0 = female; 1 = male)	−0.141	0.153	0.854	1	0.355	0.868
Subjective financial situation (reference: negative financial situation)
Above average financial situation	0.017	0.341	0.003	1	0.960	1.017
Average financial situation	0.005	0.184	0.001	1	0.978	1.005
Father has degree (0 = no; 1 = yes)	0.912	0.215	18.058	1	<0.001	2.490
Mother has degree (0 = no; 1 = yes)	0.076	0.222	0.118	1	0.731	1.079
Age (continuous variable)	0.305	0.046	44.634	1	<0.001	1.357
Type of settlement (dummy, capital city being the reference category)
County seat	0.980	0.270	13.165	1	<0.001	2.663
Small town	0.653	0.249	6.894	1	0.009	1.922
Village	1.027	0.260	15.626	1	<0.001	2.793
Economic situation of the region (0 = favorable; 1 = disadvantaged)	−0.556	0.173	10.364	1	0.001	0.573
Extramural activities
Learns in a foreign language (0 = no; 1 = yes)	0.861	0.185	21.764	1	<0.001	2.365
Reads in a foreign language (0 = no; 1 = yes)	0.348	0.205	2.874	1	0.090	1.417
Watches films in a foreign language (0 = no; 1 = yes)	0.740	0.205	13.046	1	<0.001	2.096
Keeps contact in a foreign language (0 = no; 1 = yes)	0.191	0.217	0.771	1	0.380	1.210
Constant	−8.347	0.897	86.634	1	<0.001	0.000

−2 Log likelihood: 1140.110; Cox and Snell r square: 0.136; Nagelkerke r square: 0.228.

). The explanatory variables were the father’s degree, age, county seat, small town, village, economic situation of the region, engagement in extramural English language use, learns in a foreign language, and watches films in a foreign language.

5. Discussion

Several specific factors explain the patterns of language proficiency among secondary school students in Hungary. As a post-socialist country, the languages taught and preferred have undergone significant changes in recent decades. Before the change of regime, Russian was compulsory in schools, but the effectiveness of language teaching and the population’s language skills were not at a high level. In the last decades of the socialist period, German and English were the most common second languages taught in secondary schools, with French and Italian being less common (Vágó, 2009). After the change of regime, compulsory Russian language teaching was abolished, but there were not enough qualified English and German teachers available (Nikolov, 2009), which had an impact on language teaching at the time and on the language skills of students (Vágó, 2009). In a 1994 sample, German (17.3%), English (11.5%), and Russian (8.8%) were the most widely spoken languages among adults, while among 14–17 year olds, young people reported equal levels of German and English language proficiency. At the beginning of the 1990s, German was still clearly the primary foreign language taught in schools (Vágó, 2009). By the turn of the millennium, the popularity of English had stabilized (Csizér et al., 2004; Sebestyén, 2023), while in the following decades, the number of people speaking at least one foreign language began to rise; among the adult population, the proportion of those who did not know any foreign languages decreased from 74.8% in 2007 to 57.6% in 2016, while 13.8% spoke two or more foreign languages (Central Statistical Office, 2023). Nemes and Muhariné Preszter (2019) emphasize the decline of French in public education. At the same time, the latest statistics continue to indicate that the Hungarian population’s language skills lag far behind the European Union average and those of neighboring countries (Eurostat, 2024b). Based on our research results, we can state that the gap between English and German language skills has widened in recent years, as the proportion of secondary school students reporting English language skills was 2.07 times higher in 2016, while knowledge of other foreign languages was negligible. In the last decade, the spread of English at ISCED levels 3–4 has been characteristic of most countries in the European Union, but this trend is not yet universal, even in the post-socialist region (Eurostat, 2024a). Looking at the levels of spoken languages, we can see that both English and German have risen, although basic language skills still dominate (in 2020, the figures were 54.4% and 62.8% for English and German, respectively).

The conversion of language skills into language exams increased between 2016 and 2020. However, the shift is more noticeable for German, French, Italian, and Spanish, although the low sample sizes for smaller languages call for caution. It is likely that these are mostly second languages alongside English. It is important to note that a significant proportion of students entering secondary school (2016: 15.1%; 2020: 13.6%) do not speak any foreign language, even after at least five years of language learning, according to their own statements. Based on international comparative studies, the performance of Hungarian students is greatly influenced by their parents’ socioeconomic background (PISA, 2022), and according to domestic studies, this also applies to language skills (Nikolov, 2009; Novák & Morvai, 2017). The strong social embeddedness of performance is complemented by a selective school system in which extra points earned through language exams are necessary to gain admission to higher-prestige programs. This certainly explains the growing importance of language exams for second language proficiency. The number of spoken languages showed a slight decrease between 2016 and 2020 (1.18 and 1.12), while the proportion of people speaking two languages also decreased (30.1% and 23.7%).

Based on the data obtained, we can conclude that the most common language paths among Hungarian secondary school students are Hungarian–English–German and Hungarian–German–English. Angelovska et al. (2023) pointed out that the acquisition of a second and third language (German and English in their study) is not a linear process, as language dominance is of fundamental importance and the second language (German in the study) also had a transfer effect on the third language (English). Multilingualism can therefore be interpreted as a dynamic system in the case of students as well.

We can therefore answer our first research question by saying that the shifts between the two waves of the study include positive changes (a decrease in the proportion of people who speak no languages, an increase in the level of languages spoken, and an expansion in the conversion of language skills into language exam certificates), but at the same time, the number of languages spoken has decreased slightly, and the proportion of those who know two languages is also showing a downward trend. The language skills of Hungarian high school students are characterized by the dominance of English.

In our analysis, we also wanted to examine the effects of school types, but the patterns of language proficiency are extremely unfavorable in vocational training programs that do not offer a high school diploma: we saw that only one student in the national sample had passed a language exam. The problem of effectiveness can be further illustrated by the fact that, according to European Union statistics, the proportion of students learning two languages at this school level will be negligible in 2022: only one country has a lower value (Eurostat, 2024a).

With regard to other forms of language use, we can remark that the proportions of learning did not increase significantly between the two surveys (when the “often” and “always” categories are added together, the proportions even decreased from 61.1% to 57.7%). Although 13.6% of young people stated in 2020 that they did not speak any foreign languages, the proportions in the “never” categories were significantly higher in all cases (reading online or on paper: 41.6%, watching films and series: 39%, and maintaining contact: 53.8%). In contrast, Fajt’s (2022) results show that respondents used English much more frequently in their extramural English activities: on a Likert scale of 1 to 5, the average value for watching films and series was 3.70, for reading it was 3.31, and for keeping in touch (chatting) it was 2.17. Based on the data, we can conclude that in the researcher’s study, students used English more regularly in forms related to media consumption. The differing results of the two studies may indicate that language use outside school can be influenced by several factors, including media consumption habits, which Schurz and Sundqvist (2022) also mention as a relevant factor. The effect of the mother’s degree was stronger than the father’s. The age, which was used as a continuous variable, has a significant effect too. In the case of the type of the settlement, our results do not fit into the national trends. We know the students’ results at the National Competence Assessment in foreign language (Oktatási Hivatal, 2022), and it shows a linear relationship between the type of the settlement and foreign language proficiency, but those data are based on a centralized test and not self-reported data. Moreover, the base of the National Competence Assessment is the location of the school and not the type of the student’s resident, and we focus on the type of the settlement and regional differences in our model at the same time.

We know that English has become increasingly dominant in other forms of language use over the past decade (European Commission, 2023), and research findings have shown that these activities have a positive impact on language proficiency (Webb & Chang, 2015; Peters & Webb, 2018; Huwari et al., 2023). However, we have little data on the explanatory power of these activities when we examine their effects in conjunction with sociodemographic background variables. The dependent variable in our first regression model was at least intermediate English language proficiency, which was explained by a number of factors. We found no correlation with gender or the economic situation of the region, and foreign language contact also showed no correlation. However, we found a significant correlation with above-average financial status, the mother’s degree, age, county seat, the use of foreign languages for learning, reading, and watching films. The strongest explanatory factors were reading, above-average financial status, and watching films.

The results of the study essentially confirm the conclusions of previous empirical research, according to which extramural language use—especially reading, audiovisual input (e.g., watching movies), and interactive online activities (e.g., surfing the Internet and playing video games)—significantly contribute to the development of foreign language competence (Sundqvist, 2009; Cadierno et al., 2022). We can state that the effects were stronger in the case of an informal learning process. However, in the present sample, gender differences did not show a significant effect, while the influence of social background proved to be more pronounced, which nuances the picture painted by previous results. The answer to our second research question is therefore that the explanatory variables include both sociodemographic and language-related factors.

Several factors also explain the conversion of existing knowledge into language exam certificates. As we have already mentioned, passing language exams is essential for accessing prestigious training programs, which is linked to both social and geographical mobility. Perhaps this is why the explanatory power of financial status disappears, while the father’s degree emphasizes the father’s position in the intergenerational transmission of status. This result is worth highlighting because the self-reported language proficiency was more strongly influenced by the mother’s educational level in more intense. The effect of age remains, and in our opinion, the positive explanatory power of non-capital city residence types can be explained by the intention of geographical mobility. The high-prestige training courses and institutions are located in the capital city, so the language has a key role to reach them. In disadvantaged regions, the chances of obtaining language exam certificates are lower, which is a remarkable result because the variable did not explain language proficiency. The explanation may lie in geographical differences in language teaching. However, the language exam is an objective and not a self-reported criterion. We can therefore answer our third research question by saying that the probability of obtaining a language exam certificate is also shaped by sociodemographic and language use indicators, but the significant relationships differ: we found the highest explanatory power in the case of village, county seat, and father’s degree. The significant relationship was missing in the case of “reads in a foreign language”, but it remained at learning and watching films. Moreover, the stronger relationship belongs to learning in a foreign language, which is the opposite pattern of Table 5 and highlights the conscious and planned motives in it. The variable “keeping contact in a foreign language” does not have a significant effect in either model. An important finding is that the impact of sociodemographic indicators appears to be stronger in this model, while we found no differences based on gender in either model. The explanatory power of the models suggests that the joint consideration of extramural activities and sociodemographic background variables constitutes a meaningful and relevant approach to explaining foreign language proficiency. Nevertheless, the patterns of the exploratory variables are not equivalent in the two models, so it is worth separating these fields (self-reported data on foreign language proficiency and language exam) during the later analyses. In the case of a language exam, the mechanism of social reproduction and the desire of social mobility can rather be identified as hidden motives. This may explain why the effects of financial situations disappear in this model and why the geographical patterns are so distinctive, whereas in the field of language proficiency, the informal activities appear to be more prominent (reading and watching films).

6. Conclusions

Foreign language learning is a complex process influenced by numerous factors, in which individual characteristics (age, gender, and motivation), sociocultural background (family environment and social status), and educational environment (teaching methods, classroom, and extracurricular activities) all play a decisive role. Overall, our findings highlight recent changes in the language skills of Hungarian secondary school students (the increasingly dominant role of English, rising language proficiency levels, and diverse forms of target language use), but they also draw attention to the decline in the percentage of speakers of two foreign languages, which is not a clear trend either in Europe or in the region. The issue of language teaching for students in vocational training who do not take the school-leaving exam, as well as the almost complete absence of language exams, draws attention to the problems of language teaching in this type of school, on the one hand, and to the fact that due to the specific nature of the admission system, this also reduces opportunities for social mobility, on the other. In recent years, there have been a number of changes in Hungary with regard to language exams: language exams were first made compulsory (in certain fields, they were already a condition for admission, but they were in any case necessary for graduation), but then, due to the high number of unclaimed diplomas, this regulation was abolished in 2023; however, at the time of this research, this change was not yet visible. In Hungary, language exam requirements currently vary from university program to university program, but admission to prestigious programs is virtually impossible without language proficiency verified by language exam certificates. Regression models show that both language proficiency and the existence of language exam certificates are fundamentally embedded in sociodemographic background indicators, but they are also shaped by different forms of target language use. In the case of intermediate English language skills, we can observe almost identical strength, while the types of target language use in the second model were somewhat weaker than the background variables. Passing language exams was more closely linked to goals of social and geographical mobility and maintaining status. Although social inequalities shape language proficiency and the existence of language exams, it is important to note that other ways of using the target language also had a positive explanatory power. Based on this, it would be particularly important to incorporate these activities more into the curriculum (and then into leisure time) in schools with low socioeconomic backgrounds, as access is provided through the Internet.

Our research had several limitations. Firstly, it should be noted that we based our assessment of language proficiency on subjective opinion (although this is common practice in other studies and statistics). Secondly, the questionnaire did not include an assessment of language exam levels. Due to the low number of language exam certificates in vocational training that does not lead to a high school diploma, we had to refrain from examining school types. The low number of respondents also calls for caution when interpreting the data on French, Spanish, Italian, and Russian language skills. As the regulation of language exam certificates in higher education has changed in recent years, it will be very important to assess the impact of these changes in the upcoming years, as the current secondary school population may be subject to different legal regulations. Our analysis was based on a 2020 database, so we have not yet had the opportunity to explore the latest trends (decline in the number of language exam takers and age group changes).