A study of FL children’s exposure, family backgrounds, and vocabulary learning

2.1 Extramural exposure and FL learning

A core challenge in classroom-based EFL education is the time constraint. In many EFL settings, there are usually only limited numbers of hours devoted weekly to EFL learning, as time is allocated to various other subjects such as the learners’ mother tongue and mathematics. To master any language successfully, abundant exposure is indispensable. For instance, English learners often need to know more than the most frequent 2,000 English words to be able to comprehend different texts and “there is clearly not enough classroom time to teach that many words” (Webb 2015, p. 159). As students increase their exposure to English words through extramural learning, their overall language competence is likely to improve (Lu and Dang 2023). One key reason for this is known as the frequency effect. As Ellis (2002) has reported in his seminal review article, frequency plays an important role in the processing of “phonology, phonotactics, reading, spelling, lexis, morphosyntax, formulaic language, language comprehension, grammaticality, sentence production, and syntax” (p. 143). Since language learning involves “a huge collection of memories of previously experienced utterances” (Ellis 2002, p. 166) and classroom time is limited, it can be argued that abundant out-of-class exposure is perhaps the only promising avenue for mastering a FL. Kusyk et al. (2025, p. 2) have noted in their scoping review that multiple terms have been created to refer to out-of-class and/or informal second/foreign language learning, some of which “include assigned or teacher-/other-directed activities (e.g., IDLE, LBC, LLT)”. Sundqvist (2009) introduced the term Extramural English (EE) to refer to learner-initiated exposure to English outside class. Sundqvist (2024, p. 5) clarifies that EE “does not have any connection with school or schooling, that is, no connection with formal (or “intramural”) learning”. In this article, we use the adjective extramural to indicate outside the compulsory mainstream EFL class such that informal activities (e.g., listening to songs at home), completing homework (e.g., at home/a café) and having supplementary English tutorials (e.g., at home/a cram school) are all regarded as extramural exposure to English. Note that this is different from EE Sundqvist (2009) coined and defined. We set out to focus on the English activities beyond the time spent on compulsory mainstream English class.

Studies related to young non-English-L1 learners’ extramural English exposure are limited to date, and they have mostly been conducted in Western contexts. Most researchers in these contexts have found that young learners had abundant exposure to English outside class. For example, De Wilde et al. (2020) found that at least three-quarters of their participants, 780 ten-to-twelve-year-old children in Belgium, were exposed to English daily through games, social media, and television. In Iceland, Sigurjónsdóttir and Nowenstein (2021) found similarly that about two-thirds of the ten-to-twelve-year-old participants were exposed to spoken English on a daily basis. However, the overall picture seems very different in the Asian contexts within which substantial numbers of EFL children reside. A survey carried out by Tsang (2023a) revealed that primary-school children from China, Hong Kong, Japan, Korea, and Taiwan generally spent a very small amount of time receiving spoken English outside class. Although later, Tsang and Lam (2024) found that over 90 % of Hong Kong junior-secondary-school participants had informal exposure to English outside class, the actual amount of exposure time was also very low (the estimated medians around 3–7 h monthly). These participants exposed themselves to English through seven channels: Games, videos/movies, songs, print (e.g., magazines), human-to-human communication (e.g., texting; face-to-face conversations), audio-and-print materials (e.g., audiobooks), and audio-only materials (e.g., podcasts).

2.2 Socioeconomic status (SES)

SES represents a stratifying measure that indicates a family’s income level and, in many cases, the educational attainment of either the mother, the father, or both (Pace et al. 2017). SES has been found to affect the linguistic and cognitive development of children in early childhood (Hackman and Farah 2009; Pace et al. 2017). Lee and Lee (2021) observed that lower SES students are likely to be confronted with challenges in key areas related to academic motivation and experience with technology in FL learning (e.g., access to computers). Sun et al. (2015), having adopted maternal education levels as a proxy for SES, found that EFL learners with more educated mothers would experience a greater likelihood of out-of-class English practice and vocabulary learning. In essence, learners having more resources, socially and financially, tend to have more opportunities to access EFL resources outside class (Hu 2009). Poorer and/or less educated parents tend to lack the means to educate and invest themselves and/or their children in the FL; they often have difficulty in supporting their children’s FL learning and exposure in extramural settings as well as higher SES families can. Forey et al. (2016) reported some parents in Hong Kong feeling insecure and a lack of confidence about involving in their children’s English learning such as through reading aloud to them. This indicates that parents’ own competence (and somewhat indirectly educational levels) can play a role in children’s success in FL learning.

Although many educational researchers take SES into account, most have incorporated only one or two measures (e.g., only maternal education) in their studies. Examining a wider range of SES variables would yield a more granular understanding of the relationships between SES and extramural English learning, thereby allowing more nuanced conclusions and constructive implications to be drawn. We set out to address this gap in the current study.

2.3 FL vocabulary

Unsurprisingly, previous studies have reported a generally positive relationship between extramural English exposure and proficiency levels among young FL learners (e.g., Leona et al. 2021; Lindgren and Muñoz 2013; Puimège and Peters 2019; Sylvén and Sundqvist 2012; Tsang 2025). Many researchers investigated vocabulary as a proxy for proficiency. Puimège and Peters (2019), for example, had 560 Dutch-speaking children in Belgium complete vocabulary tests and a questionnaire. Although these children had not begun to receive formal English education at that time, a positive relationship was found between English gaming and TV viewing and learners’ performances on an English vocabulary test. Leona et al. (2021) recruited 298 fourth-grade Dutch primary-school children in the Netherlands in their study. Using questionnaires and vocabulary tests, they found that extramural English engagement via entertaining media (e.g., films) and family (e.g., speaking English at home) were, overall, positively related to vocabulary test scores. These positive findings make good sense as extramural English exposure dovetails some important principles for successful vocabulary learning, such as receiving large amounts of input and having opportunities for repetition and noticing (Coxhead and Bytheway 2015). This small body of research has mostly been conducted in European contexts, yet very few relevant studies have been conducted in Asia. Also, while researchers have employed various validated vocabulary tests in their studies (e.g., Peabody Picture Vocabulary Test in Leona et al. 2021), there is a general lack of examination of learners’ mastery or gains in vocabulary at different frequency levels. These levels are commonly categorized as the most frequent 1,000 words, the most frequent 1,001–2,000 words, and so forth (e.g., Webb et al. 2017). It is important to examine learners’ mastery of different levels closely such that specific implications can be drawn for effective vocabulary learning and teaching (e.g., medium of instruction and mastery of different frequency levels in Lo and Murphy 2010). The current study therefore set out to fill this gap in a bid to illuminate a more nuanced understanding of the relationship between vocabulary learning and extramural English exposure.

4.1 Participants

Ninety Grade-5 EFL children (M _age  = 9.988, SD = 0.570; Males = 46, Females = 44) studying at three primary schools in Hong Kong participated in the study voluntarily. Although located in different districts in Hong Kong, the three participating schools offered typical EFL education to native-Chinese-speaking children from Grade 1 onwards. They could therefore be generally considered as typical EFL learners in the context of Hong Kong. As shown by the absence of between-school significant differences in both the pre- and post-test scores, the participants from the three schools had comparable vocabulary sizes at both T1 and T2 of the study. Ethical approval was granted by the institution with which the first author is affiliated, and consent to participate was sought from the participating children and their parents/guardians.

The children participants, together with their parents/guardians, were invited to participate in an individual interview after the post-test was completed. In total, 90 face-to-face interviews were conducted by trained assistants. The interviewees were 77 mothers, seven fathers and one aunt (a guardian), and there were five sessions in which both fathers and mothers attended. The assistants first checked to establish the interviewees’ eligibility (i.e., they had a sound knowledge of the children’s English exposure from T1 onwards). The assistants invited ineligible parents/guardians to recommend another family member/guardian who had the required knowledge to take part in the interview instead. The children were also encouraged strongly to participate in the interviews as far as possible and they were present in 82 of the 90 interviews.

4.2 Measures and procedures

4.2.1 EFL exposure

The children’s exposure to English from T1 to T2 was established from the individual structured interviews. The regularities in these young children’s habits and timetables, and the parents’/guardian’s long-term presence immensely facilitated the recall of events. In each session, the trained assistant, the parent(s)/guardian, and the child (in 82/90 interviews), collaborated to co-construct all English exposure beyond the regular EFL lessons the child had received in the past year. To better structure the recall, the assistants focused first on supplementary classes (i.e., English tutorial classes; homework completion classes; see Table 1 for definitions), then English homework, and finally, other exposure such as self-initiated/parent-assigned tasks. To assist the interviewees in their recall, the interviewers guided them to consider two aspects: Spoken (i.e., listening and speaking) and written (i.e., reading and writing) English. The assistants would ask for clarification and examples, and noted the quality (i.e., type) and quantity (i.e., amount of time) of exposure on a log sheet (Figure 1) throughout each interview. At the end of each interview, the assistant summarized the profile for the interviewees as a form of member check. This bottom-up qualitative approach yielded a holistic and relatively accurate picture of a child’s total English exposure beyond the mainstream EFL class over a 1-year period. Each interview lasted around 20–60 min, depending largely on the interviewees’ responses (e.g., an interview would end quickly when the child had little to no out-of-class exposure). All the interviews were conducted in Chinese (all interviewees’ and interviewers’ first language). For more details about the interview and data collection in relation to EFL exposure, see Tsang 2023c).

Table 1:

Definitions and examples of English exposure the participants had from T1 to T2.

Category	Definition/elaboration	Examples
English tutorial classes	Supplementary (usually fee-paying) classes that primarily aim to increase learners’ English proficiency. Their focus can be narrow (e.g., grammar; writing) or broad (e.g., all four macro-language skills).	One-to-one private tutorial sessions at home; small-group classes in tutorial schools
Homework completion classes	Supplementary (usually fee-paying) classes in which tutors assist learners to complete their homework assigned by teachers in mainstream schools. When the homework is completed, the children are usually asked to do revision or additional exercises.	One-to-one private sessions at home; small-group classes in tutorial schools
English homework	Tasks assigned by English teachers to do outside class	Exercise sheets; writing tasks
General informal exposure (GIE)	This represents informal engagement with English for entertainment, learning, or communicative purposes. The subcategories are shown below.	–
– Videos	Receiving audio-visual input with or without written English input (e.g., captions)	Cartoons; movies
– Print	Receiving written English input	Books; online texts
– Songs	Receiving spoken English input alongside melodies	Children songs; pop music
– Oral communication	Situations where learners engage in oral communication using English.	Conversing with a non-Chinese-speaking domestic helper; talking with native English speakers
– Games	Receiving spoken and/or written English input while playing games	Mobile games; games on Nintendo switch
– Applications	English learning applications available on phones and/or computers	Duolingo
– Drama	Only one participant engaged in drama performance activities in English.	–

Figure 1:

The interview log sheet.

4.2.3 Vocabulary size

The updated Vocabulary Levels Test (Webb et al. 2017; accessible directly via Paul Nation’s resources website, Nation n.d.) was administered to the children. The first three levels, namely 1,000, 2,000, and 3,000 word levels, were administered. As agreed by the first author and the participants’ English teachers, these tests covered a wide array of words from very simple (e.g., eye and box in level 1,000) to difficult ones (e.g., enthusiastic and approximate in level 3,000) for the target participants, hence highly appropriate for use, especially in a longitudinal study. There were 30 items in each level, yielding a total of 90. This passive vocabulary test was slightly re-formatted and colored to make the task clearer for the children, who were told that the words or phrases in yellow were definitions (see Figure 2). They were required to tick one of the words in green that matched each definition. The test, therefore, essentially measured passive vocabulary size. All the participants in the same school completed the test simultaneously in classrooms on Saturdays. No time limit was given and the participants were told explicitly and repeatedly that they were not allowed to guess any answers (as doing so would affect the accurate measurement of their genuine vocabulary sizes). The tests were checked whenever possible as the administrators collected them (e.g., verbally checking their understanding of a few words in green that children ticked, to make sure they were not ticking the boxes randomly).

Figure 2:

Part of the vocabulary test administered in the study (adapted from Webb et al. 2017).

4.3 Data entry and analysis

The exposure information on the interview log sheets was first entered directly into a spreadsheet. The types of exposure were then categorized using a bottom-up approach (e.g., watching cartoons, Netflix, and YouTube all grouped under ‘videos’). The complete list of exposure categories, together with definitions and examples, is shown in Table 1. The total amount of time each individual spent on each category was summed for later analysis. As for the indicators of SES, following Lau and Cheng (2016), each parent’s highest education level was categorized into one of the following five levels: (1) junior high school or below, (2) senior high school, (3) pre-degree, (4) bachelor’s, and (5) master’s or above. Their household income, which was in Hong Kong dollars, was entered directly into the spreadsheet. For the vocabulary tests, one mark was awarded for each correct answer, yielding a total of 30 marks per level. Tests not completed seriously (e.g., ticking all the boxes in the same column in the entire test) were discarded. Each test was marked by one assistant and cross-checked by another. Discrepancies were discussed. As the current study focused on vocabulary gains from T1 to T2, tests showing a total score difference (i.e., T2–T1) of 0 or below were not analyzed. Tests with any negative score difference at 1,000, 2,000, or 3,000 levels were also discarded from this study.

To address RQ1, descriptive statistics were used. As the exposure data did not generally follow normality, non-parametric tests (Spearman’s correlations and the Mann-Whitney U test) were used to address the relationships under investigation in RQ2 and RQ3. To shed further light on how exposure could potentially be related to vocabulary gains, the profiles of the five most improved participants were also analyzed and are reported below.

5.1 The participants’ exposure to English (RQ1)

Table 1 shows the categories, together with their definitions/elaboration and examples, of all types of exposure the children had outside the mainstream EFL class over a one-year period. There were two types of common academic classes, English tutorial and homework completion classes. The interviewees also reported a range of self-initiated or parent-assigned activities through which their children were exposed to English.

Table 2 presents the descriptive statistics for the various kinds of English exposure the participants engaged with from T1 to T2. Academically, around 72.222 % and 28.889 % of the participants took tutorial and homework classes respectively. Everyone (except for one missing datum) had experience with homework. 75.556 % engaged in informal activities, the most frequent being videos (52.222 %), followed by print (36.667 %), songs (13.333 %), and oral communication (12.222 %). Fewer than 10 participants were exposed to English through games, applications, or drama. The skewness and kurtosis values indicate that the data were generally not normally distributed.

5.2 Correlations between parents’ education levels, household income, and exposure (RQ2)

As shown in Table 3, the medians of the highest qualifications for both parents were 2 (senior high school). The average household income was around HKD35000 but there was a large variation, as indicated in the standard deviation (SD = HKD23219.246). These background variables did not seem to be significantly correlated with most kinds of exposure. However, income was significantly correlated with GIE, r _s  = 0.225, p = 0.038. Videos were the only type of exposure found to be positively correlated with household income, r _s  = 0.356, p = 0.017. If accepting a slightly higher Type-1 error rate of 10 %, to make up for the low power in this small sample size, GIE was also potentially related to both father’s (r _s  = 0.206, p = 0.064) and mother’s education (r _s  = 0.176, p = 0.097), with a small effect.

To compare if parents’ education levels and household income were significantly different between the ‘Yes’ and ‘No’ exposure groups, the Mann-Whitney U tests were administered. No significant differences were found in any of the comparisons (see Table A in the Appendix) except for print and father’s education, U = 553, p = 0.019, which had a small to medium effect, r = −0.259 (Field 2018). Figure 3 presents the participants' exposure to print categorized by paternal education.

Figure 3:

The participants’ exposure to print categorized by paternal education.

5.3 T1 and T2 vocabulary scores (RQ3)

Table 4 shows the descriptive statistics for the pre- and post-vocabulary test scores for all participants. The average scores at T2 were higher than the corresponding scores at T1. Throughout the year, the participants, on average, gained around 2–3 words at the 1,000 level and about 2 words at the 2,000 level, but the gain at the 3,000 level was minimal (0.658 words only). Overall, the participants gained slightly more than an average of 5 words from T1 to T2.

The present study focused on vocabulary development. Therefore, those who did not show improvement or who did worse were removed from further analyses. The cleansed results are shown in Table 5. This group of learners gained around 3–5 words at each level and approximately 8 words in total. As a reviewer aptly pointed out, the genuine gains in these learners’ vocabulary size would likely be much larger. For instance, if to extrapolate, gaining 3 out of 30 words at a level equates to 10 % of the words at that level. The paired-sample t-tests results are also shown in Table 5. All three levels and the total scores showed that the differences were statistically significant (all ps < 0.001) and the effect sizes were large (all ds > 1).

To find out the relationships between the gains in scores and various kinds of exposure, Spearman correlations were performed. The results are presented in Table 6. No statistical difference was found at levels 1,000 or 3,000. GIE was the only area showing a small to medium significant correlation with gains in vocabulary at level 2,000 (r _s  = 0.330, p = 0.031) and overall vocabulary (r _s  = 0.320, p = 0.013). Interestingly, time spent on tutorial classes and homework were found to be associated negatively with total vocabulary gains (r _s  = −0.273, p = 0.036 for tutorial classes; r _s  = −0.291, p = 0.025 for homework).

The profiles of the five most improved students are illustrated in Table 7. These students gained 19 to 26 words from T1 to T2. All of them had tutorial classes during the period although the amount of such tuition was not high (from 4 to 34.667 h only throughout the whole year). Only #5 attended homework classes. #1–#4 had substantial amounts of engagement with GIE, placing them above the 90th percentile (786.5 h) among all the participants. All five of them were exposed to English through videos and four of them also had exposure to oral communication.