Factors Affecting Home Language Literacy Development in Japanese-English Bicultural Children in Japan

Abstract

Simultaneous bilingual and bicultural children who are schooled in the dominant societal language can acquire literacy in their home language through home literacy practices and weekend school. Twenty-eight Japanese-English bilingual–bicultural children (ages 9 to 14) attending English weekend schools in Japan were assessed using the standardized Test of Written Language (TOWL, 4th ed.). Their overall age-adjusted writing scores showed that most were on a par with same-age US peers. The parents’ responses to parental questionnaires revealed that the children’s regular attendance and enjoyment of weekend school, coupled with parental home support and access to plenty of appropriate English reading materials, helped to establish their English literacy. Nevertheless, the children attained different writing scores. Although many performed within the Average range, some scored either in the Above Average or higher range or the Below Average or lower range. Further analysis of these three groups of children revealed a strong relationship between reading practices and writing scores. Better writers tend to read frequently in English; they enjoy reading independently and voluntarily. These findings underscore the importance of recreational reading as an accessible and affordable means for developing home language literacy in bilingual and bicultural children in Japan.

1. Introduction

Simultaneous bilingual children are well-positioned to acquire literacy and cognitive academic language proficiency (CALP, Cummins 2000) in their two languages because their oral ability or basic interpersonal communication skills (BICS) can help them to establish stronger connections with corresponding written forms (Hornberger 2003). Such children can develop biliteracy even when schooled only in the societal language if they learn a home language outside school, as extensive research cited by Cummins (2021, cf. Chapters 3 and 9) supports the crosslinguistic interdependence of literacy-related knowledge and skills that hold even across typologically different languages. When both parents read to their children from birth in their respective languages in a “one parent, one language” (OPOL) setting, biliteracy may even begin in infancy. Thus, bicultural children in mixed-heritage families have the potential to acquire literacy in two languages from a very young age.

In Japan, the number of bicultural children is increasing due to more marriages between Japanese and non-Japanese nationals in recent decades. About one in fifty babies that were born annually in Japan between 2000 and 2021 had a non-Japanese parent (Ministry of Health, Labour and Welfare 2023). However, not many bicultural children grow up to become bilingual and biliterate. Receptive bilingualism is common, even when the home language is a prestigious foreign language like English (Noguchi 2001). The pervasiveness of the societal language steers children towards receptive bilingualism or even Japanese monolingualism, with biliteracy being an even more unattainable goal (Yamamoto 2001). Japan’s Ministry of Education, Culture, Sports, Science and Technology (MEXT) offers no official policy for home language education in mainstream education, so there is minimal or no support in public schools (Majima and Sakurai 2021). Although English has become a foreign language subject from Grade 5 in Japanese elementary school, children whose home language is English do not benefit much from it. According to national curriculum standards (MEXT 2022), Japanese monolingual fifth and sixth graders learn to identify letters and understand the meanings of simple English words/phrases, but such lessons are too easy for bicultural children who may be reading chapter books in English at those grades. Attending an international or minority language school is one way to develop home language literacy, but it is not an option for every child. Therefore, parents would need to teach their children to read and write in their home language, but many do not have the time and energy to do so.

Despite the prestigious status of English as a foreign language in Japan, it is not easy for bilingual children to develop Japanese-English biliteracy because they have to learn two different writing systems. Japanese has three different types of script (the syllabic katakana and hiragana and the logographic kanji characters) versus the English alphabet. It is particularly difficult to develop English writing skills than reading skills because more explicit instruction (e.g., copying words) is required than, for example, bedtime reading. Writing is a more demanding task than reading; it comes later in the language learning process and takes longer to master (Shanahan 2006). The complexity of the writing process makes writing even harder to acquire at home. Therefore, although well-placed to become biliterate, not many simultaneous Japanese-English bilingual and bicultural children in Japan develop high English literacy skills at home. This study investigates a specific group of children who developed literacy in their home language despite these challenges by learning English informally at home and in “weekend schools” that provide them with level-appropriate English literacy instruction that is unavailable in Japanese public schools.

1.1. The Home Environment: Parental Roles and Literacy Resources

When home language instruction is lacking in regular schools, the family has to play a central role in the child’s home language literacy development. However, exposure at home only develops oral skills; it does not provide enough support for literacy development (Kang 2015). Regular literacy practices are necessary, particularly as children become older and increasingly dominant in their school language (Oriyama 2016). Such practices can expose children to richer and more varied linguistic input that is unavailable in routine family conversations. Even when children do not speak the home language, literacy instruction helps to develop linguistic competence and a positive attitude toward it (Smith-Christmas 2016).

However, we know little about the effectiveness of home literacy practices, probably because few bilingual families embark upon or continue down the path to biliteracy. Compared to families where both parents speak the home language, mixed-heritage families are less likely to teach home language literacy (Karpava 2021). Okita (2002) only found two of twenty-eight Japanese-English intermarried families in the UK who persisted with Japanese literacy learning at home until their children became teenagers. Parents who continue literacy efforts likely have positive attitudes toward the home language (Fu 2020; Lao 2004) and strong impact beliefs (De Houwer 1999), i.e., a conviction that they could affect their children’s language learning. Parents who are teachers, moreover, may have strong impact beliefs about teaching the home language (Fu 2020), most likely because they have the skills to teach their children. Peer influence can also strengthen impact beliefs about biliteracy, as occurred for an English-speaking parent in Japan who decided to teach her child to read after seeing other English-speaking parents doing so (Nakamura 2019).

Parents’ attitudes and impact beliefs translate into efforts such as shared reading and reenacting school “literacy” activities through play (Reyes 2012). Bosma and Blom (2020) found that such activities at home are more important for home language development than other activities like watching television. Since parent–child book reading creates a literacy-rich home environment from early on (Collins and Svensson 2008), it is unsurprising that shared reading in a home language also helps to develop the school language for bilingual children (Willard et al. 2021). Parents may go to great lengths to help develop their children’s literacy by making their own teaching materials (Saunders 1982) and using creative methods of instruction, e.g., drama, puppets and drawings (Kopeliovich 2013). They may read lengthy books to expose children to more complex texts and continue doing so even when their children are older and can read independently (Nakamura 2019).

Parents also prepare the home literacy setting for the child. The number of books in the home language is linked to children’s vocabulary development and use of that language (e.g., Rydland and Grøver 2021). The choice of books also matters; only books that match children’s developmental needs and actively engage them in reading or writing translate into successful literacy learning (Li 2007). Reading for pleasure develops reading skills in the home language (Taniguchi 2021). While we know less about writing, reading skills predict writing ability (e.g., Kim et al. 2018), as both share cognitive processes and knowledge bases (Fitzgerald and Shanahan 2000). This is confirmed by Kim and Pyun’s (2014) finding that the frequency of Korean reading and writing practice predicted the writing proficiency of Korean home language learners.

1.2. The Role of Weekend Schools in Biliteracy Development

Bilingual children can attend weekend schools to acquire literacy skills in their home language when such instruction is unavailable in mainstream education. Weekend schools (also called “complementary schools” in the UK, “community schools” in Australia, and “heritage language schools” in the US) are voluntary establishments that perform several functions. While focusing on language instruction, particularly reading and writing, they also teach the minority culture through language, crafts, music, or dance (e.g., Kenner 2004; Li and Wu 2010) and serve as a community space for families (Ganassin 2020). These activities facilitate children’s language and identity development (e.g., Prokopiou and Cline 2010). Although weekend schools face many challenges, such as operating in borrowed spaces, a lack of resources, and limited instruction time (Lee and Chen-Wu 2021; Nordstrom and Jung 2022), they are necessary for home language literacy learning. Children who attend weekend school have higher literacy skills in their home language compared to those who do not attend (Lao and Lee 2009; Mattheoudakis et al. 2020). They also tend to do better at their regular school (Barradas 2004). In the UK, weekend schoolers demonstrate a more positive attitude toward education—they are more attentive, well-behaved, and motivated to learn (Department for Children, Schools and Families 2010).

In Japan, home language instruction is lacking in the public education system. Some schools use the children’s home languages to teach academic subjects and promote inclusivity, but they are not taught as language subjects per se (Majima and Sakurai 2021). Bogo kyooshitsu, or “mother tongue classrooms”, have been established since the 1990s to help minority children learn their home language (Matsubara 2004). Mainly run by volunteers in community centers on weekends, they usually teach the languages of Japan’s largest immigrant groups (e.g., Chinese, Korean, Spanish, Vietnamese, and Portuguese). How such classes are run depends on the educators’ and parents’ expectations and the children’s proficiency levels. Their effectiveness is hampered, however, by the lack of teacher training and teaching resources, a dependency on volunteers, limited instruction times, the children’s mixed abilities, and a lack of motivation to learn the language (Saito 2005).

Distinct from the volunteer-run bogo kyooshitsu is the parent-run weekend school. The number of such schools is unknown; those reported within research teach English, but weekend schools in other languages also exist. Parents teach or hire teachers for lessons to be taught at one of the families’ homes or an external venue (Pauly and Yamane 1999). Unlike the bogo kyooshitsu attended by mostly sequential bilingual children, English weekend schoolers are mainly bicultural children with one parent from an English-speaking country, although English-speaking parents of other nationalities also enroll their children. Nakamura’s (2019) ethnographic study of eight parents whose Japanese-English bilingual children attended a weekend school in Tokyo showed that they believed that a high level of English literacy would help their children be accepted into better schools, including high schools or colleges abroad. They also had a strong impact belief that they could develop their children’s English literacy with support from their spouses, other parents, and the weekend school. Their language ideology and strong impact belief encouraged regular home literacy practices such as reading aloud to their children and supervising homework.

1.3. The Context for the Present Study

Studies of biliteracy generally focus on early elementary rather than late elementary and middle-school children, and sequential rather than simultaneous bilinguals (see Rothou and Tsimpli 2020, for a review). Biliteracy research on older simultaneous bilingual children schooled in the societal language is needed because they likely become increasingly dominant in that language as they grow older. Taura and Taura (2012) is a rare 14-year longitudinal case study of a Japanese-English simultaneous bilingual child who was mostly educated in Japan from Grades 1 to 12. Despite her predominantly Japanese-medium education, she performed at almost age-appropriate levels in standardized oral and written tests in English from age 11 onwards after an 8-month stay in Australia and subsequent enrollment in a private bilingual high school in Japan. This made the researchers wonder “whether intensive exposure at a certain age, on top of constant exposure since birth, is essential to boost a bilingual child’s non-dominant language to the monolingual level” (p. 501). Without such intensive exposure, regular home literacy practices and weekend school classes are necessary (Mattheoudakis et al. 2020). Therefore, in this study, we investigate how these two factors may contribute to different levels of home language literacy. We focus on older children to understand how biliteracy further develops in late childhood and adolescence.

There is also a paucity of systematic research on the short-term and long-term effects of weekend schooling (Lee and Chen-Wu 2021). Specifically, language assessment practices are hardly known (Kondo-Brown 2021). Previous research has often relied on self or parental assessments of language proficiency (Lao and Lee 2009; Mattheoudakis et al. 2020). Only Shen and Jiang (2021) adopted standardized proficiency tests for non-native speakers to measure the Chinese ability of Chinese-Australian children in Australia. The current study fills a gap in the research by objectively measuring the English writing ability of Japanese-English simultaneous bilingual and bicultural children using a standardized writing assessment. We ask the following: (1) how well can Japanese-English bicultural children attending English weekend schools write in English? From the data collected on family background and language practices, we also ask: (2) how do home and weekend school factors differ among stronger and weaker writers?

2. Materials and Methods

2.1. The Weekend Schools

We recruited participants from two English weekend schools in Tokyo and Yokohama. Both schools have been running out of community centers for over ten years. Approximately 40 children are enrolled in four or five classes of different levels in each school. The children have one hour of instruction every Saturday, but older children in Tokyo receive 1.5 h in higher-level classes. Both schools have three terms each year, comprising 34 weeks in Tokyo and 36 in Yokohama, with the annual class time being either 34 (or 51 for older children in Tokyo) or 36 h. However, hardly any child attends all the classes as they usually have regular school on Saturdays once a month, so most children have only about 30 h of class time annually. During the COVID-19 pandemic, both schools shifted to a synchronous online format for over two years before reverting to their original in-person classes. However, the data for this paper were mainly collected before online lessons started (see Nakamura and Quay 2023, for longitudinal results on weekend school learning during COVID-19).

Qualified native English-speaking teachers with experience in international or Japanese schools teach at both schools. The literacy-focused lessons include reading (e.g., phonics), discussions of written texts, vocabulary activities, and writing tasks to acquire proper spelling, punctuation, and grammar. The Tokyo school adopts a content-based approach with a specific language arts, science, or social studies theme for each term using US language arts materials. The Yokohama school takes a communicative approach to engage the children, i.e., games and activities (e.g., running dictation). Both schools provide homework so that the children continue reading and writing on weekdays. Some homework is completed online with electronic books and vocabulary/spelling games. Children in higher-level classes undertake writing projects where they plan, write, and revise an essay. Teachers assess the children periodically for class placement or promotion to a higher-level class and check homework. However, no formal year-to-year assessment exists. Some older children sit for external English examinations, but not all have had their English formally evaluated.

2.2. The Participants

This paper reports on 28 bilingual–bicultural children (16 boys and 12 girls) whose profiles are shown in the left column of

Table 1. Profile of the children and their non-Japanese parents.

Children’s Profile		No.	Total	Non-Japanese Parents’ Profile		No.	Total
Gender	Male	16		Country	US	13
	Female	12	28		UK	3
Age (9–14)	Median	10			Australia	3
Grade (4–9)	Median	5			Canada	2
Non-Japanese parent	Mother	15			Indonesia	1
	Father	13	28		Malaysia	1	23
Birth order	First	22		Years in Japan (3–30)	Mean	15.5
	Second	6	28		SD	5.97
Siblings	None	8		Age (36–56)	Median	42
	One	17		Education
	Two	3	28		Bachelor	14
Years in weekend schools (1.5–10)	Mean	5.4			Master	7
	SD	1.8			Doctorate	2	23

. Their age ranges from 9 to 14, with a median age of 10. They are in Grades 4 to 9 with Grade 5 as the median. All children have one non-Japanese parent, 15 with a non-Japanese mother and 13 with a non-Japanese father. Most children are first-borns (n = 22, 78.6%). Eight are only children, 17 have one sibling, and three have two. The children’s average length of weekend school attendance is 5.4 years, with the shortest being 1.5 years and the longest being 10 years.

The right column of Table 1 shows their non-Japanese parents’ profiles. There are only 23 parents versus 28 children because ten are siblings from five families. Many of the children’s non-Japanese parents are from the US (n = 13), but some come from the UK, Australia, or Canada. Two siblings have an Indonesian parent and one child a Malaysian parent, but these two parents have high English proficiency. The non-Japanese parents have lived in Japan for 3 to 30 years, with a mean of 15.5 years. Their ages range from 36 to 56, with a median of 42. All non-Japanese parents are highly educated, as fourteen have a bachelor’s degree, seven have a master’s degree, and two have a doctoral degree. Many have English-related jobs, e.g., teaching, translating, writing, or editing. Many Japanese parents also have an undergraduate degree or higher, but a few only completed high school or junior college.

The children were born in Japan except for US-born LET (the pseudonyms used for each participant are shown later in Table 2), who learned to read and write Japanese at a Japanese weekend school in the US before coming to Japan at age six. SAK attended a year of kindergarten in the US, as did LIS for two months. AMS also lived in the UK from ages 3 to 6. Nevertheless, all the children have attended public schools in Japan since Grade 1. For this reason, and the fact that they have received exposure to Japanese from birth from their Japanese parent, we do not assess their Japanese literacy, but confirmed through their school grades that they all have Japanese grade-level abilities. The children have received level-appropriate English instruction, not from their regular Japanese school but from their weekend school. As mentioned earlier, school English lessons do not advance the literacy abilities of bilingual children who speak English at home and have been learning to read and write from a young age. For example, Japanese monolingual school children may be learning to write alphabets and simple words when our bilingual participants are writing stories.

2.3. Assessment and Analyses of Data

We used the Test of Written Language or TOWL (4th ed., Hammill and Larsen 2009), a standardized test designed for US children aged 9 to 17, to assess our bilingual children’s English writing ability. The TOWL examines “contrived” and “spontaneous” writing across seven sub-tests. “Contrived” writing tests discrete aspects of written discourse such as vocabulary, spelling and punctuation, logical sentences, and sentence combining (see Hammill and Larsen 2009, for sub-test details). The child writes a story in response to a stimulus picture in “spontaneous” writing. It tests children’s use of contextual conventions (e.g., paragraph use and sentence construction) and story composition (e.g., the storyline and sequence of events). The TOWL is based on a large normative sample of 2205 children from 17 states in the US. While a monolingual assessment tool may be inappropriate for emerging bilinguals who are learning to speak and read a second language (Butvilofsky et al. 2021), it can be an objective assessment tool for simultaneous bilinguals who have been reading and writing in that language from a young age. Designed for children and teenagers, the TOWL is more suitable for our child participants than English tests meant for L2 adult learners. The TOWL allows us to derive age-adjusted scores for our participants of different ages and levels, and to determine their strengths and weaknesses in different aspects of English writing.

The assessments were conducted strictly according to the instructions in the TOWL examiner’s manual. Before this research, we underwent training, as prescribed by the examiner’s manual, by practicing the scoring of ten sample essays, calibrating our scores to that of the scoring keys, and conducting trial assessments on four subjects. Our participants took the 90-min assessment individually with one of us before or after their weekend school lessons. During the COVID-19 pandemic, we switched to online testing for nine children. We graded each child’s writing separately according to the guidelines in the examiner’s manual. We then compared scores for all 136 items on the test and discussed differences before arriving at the final scores for each child. Inter-rater discourse (as advocated by Matthews 2023) is thus an important component of our study for ensuring reliable assessment results.

In using the normative scores, with seven descriptive terms ranging from Very Poor to Very Superior, our participants’ writing performance is benchmarked against same-aged US children (e.g., the descriptive term Average indicates that a participant is performing in this range as established for a same-age US child). However, these scores and their descriptive terms were not used for comparing our participants with US children, but for determining their level of English writing ability and for identifying stronger and weaker writers among our participants. Specifically, the scores are used to place the children into three groups: those who perform at the Below Average range or lower, those who perform at the Average range, and those who perform at the Above Average range or higher. The children are grouped in this way so that we can understand how specific home and weekend school factors relate to different levels of writing performance.

To supplement the TOWL assessments, we also collected information about the families and their home language practices from a two-part parental questionnaire. The first part asks about the family’s background information and their language history (see Appendix A for the questions used in this paper). The second part requires parents to choose a response to 52 seven-point Likert-type items. Twenty-two items related to home and weekend school practices were analyzed for this paper (see Appendix B). The remaining questionnaire items (not shown) are analyzed and discussed elsewhere. The design of the questionnaire is based on the findings of an earlier ethnographic study on the Tokyo weekend school, where eight parents, including three from the current study, were interviewed about their home language and literacy practices (Nakamura 2019). This questionnaire was piloted on four families and revised based on the feedback received. In this study, non-Japanese parents were asked to complete the questionnaire for each child; one exception is the Japanese mother of ARS (Table 2, no. 18), who completed the questionnaire for ARS because her American husband was unavailable. The Japanese mother, however, is an English conversation teacher.

3. Results

3.1. Overall Writing Scores and Language Use at Home

Table 2 shows the TOWL results, preceded by the parental language use at home (see Appendix A, nos. 3, 4, and 5 of the questionnaire) and the children’s English use (see Appendix B, no. 23). The children’s general writing performance in the penultimate column is indicated by their overall writing composite index score, which is based on a mean of 100 (see Quay and Nakamura 2022 for detailed results of each sub-test). The composite index score is age-adjusted, e.g., a nine-year-old’s score is calculated based on the norms for US nine-year-old children. A descriptive term is assigned to each band score in the final column: Very Superior (>130), Superior (121–130), Above Average (111–120), Average (90–110), Below Average (80–89), Poor (70–79), and Very Poor (<70). The children’s overall writing scores (shown in descending order) indicate three different groups of writers. The first six children scored above 111, either Above Average or Superior (nos. 1 to 6). The next 18 children (nos. 7 to 24) scored between 90 and 110, or Average. The last four children (nos. 25 to 28) scored less than 90, either Below Average, Poor, or Very Poor.

The questionnaire asks parents to describe their language input to their children at four different points in time, i.e., currently, before elementary school, before preschool, and at birth (see Appendix A, nos. 3 and 4). Quay and Nakamura (2022) analyzed changes in language exposure over time and found that most children received consistent exposure to English from birth. Therefore, we only analyze the parents’ and children’s current language exposure in the present paper. While 15 children are exposed to the OPOL approach, 13 heard more English (More ENG) at home because their Japanese parent (JP) also used some English. ENG-JPN indicates mostly English with some Japanese, while JPN-ENG is mostly Japanese with some English. Six non-Japanese parents (NJPs) also used some Japanese (ENG-JPN). The home language exposure, particularly the additional English input from Japanese parents, contributed to the children’s active bilingualism. Twenty-one children spoke English at home more than 90% of the time, six 60–75% or 75–90% of the time, and only one 30–45% of the time. Note that, despite consistent English exposure through OPOL and responding mostly in English when addressed in that language, four children (nos. 25 to 28) obtained writing scores in the Below Average to Very Poor range. Twelve of eighteen Average performers (66.7%) received More-ENG exposure, whereas five of six children (83.3%) who performed at Above Average or higher were exposed to English in an OPOL setting. This shows that children who received English input in the More-ENG setting did not write better than those who received English input in the OPOL setting.

Table 2. TOWL results and language use in the family.

	Child	Sex	Age	NJP’s Language Use	JP’s Language Use	Current Home Exposure	Current Child ENG Use	Score	Descriptive Terms
1.	LIS	F	12	ENG	JPN	OPOL	>90%	124	Superior
2.	SAK	F	10	ENG-JPN	JPN	More ENG	>90%	123	Superior
3.	YOD	M	14	ENG	JPN	OPOL	>90%	123	Superior
4.	LUH	M	9	ENG	JPN	OPOL	>90%	119	Above Average
5.	AYO	F	12	ENG	JPN	OPOL	>90%	118	Above Average
6.	HUG	M	9	ENG	JPN	OPOL	>90%	115	Above Average
7.	EMS	F	9	ENG	JPN	OPOL	>90%	109	Average
8.	KAT	M	9	ENG	ENG-JPN	More ENG	>90%	109	Average
9.	KOO	F	9	ENG	JPN-ENG	More ENG	>90%	106	Average
10.	TAK	M	9	ENG	JPN	OPOL	>90%	104	Average
11.	MAS	F	11	ENG	ENG-JPN	More ENG	>90%	103	Average
12.	AMS	F	11	ENG	JPN-ENG	More ENG	>90%	102	Average
13.	KAN	F	10	ENG-JPN	JPN-ENG	More ENG	75–90%	102	Average
14.	LUM	M	9	ENG-JPN	JPN-ENG	More ENG	60–75%	101	Average
15.	JAS	M	10	ENG	JPN	OPOL	>90%	100	Average
16.	ERS	F	10	ENG	JPN-ENG	More ENG	>90%	96	Average
17.	MOH	M	11	ENG	ENG-JPN	More ENG	>90%	95	Average
18.	ARS	F	12	ENG	JPN-ENG	More ENG	30–45%	95	Average
19.	YUK	M	9	ENG-JPN	JPN-ENG	More ENG	75–90%	95	Average
20.	TYA	M	12	ENG	JPN-ENG	More ENG	>90%	94	Average
21.	LET	M	9	ENG	JPN	OPOL	>90%	94	Average
22.	ISN	M	12	ENG	JPN	OPOL	60–75%	92	Average
23.	KYM	F	12	ENG-JPN	JPN-ENG	More ENG	75–90%	91	Average
24.	CAS	M	14	ENG	JPN	OPOL	>90%	90	Average
25.	ALN	M	12	ENG	JPN	OPOL	>90%	86	Below Average
26.	SHS	M	10	ENG-JPN	JPN-ENG	OPOL	75–90%	83	Below Average
27.	JOG	M	9	ENG	JPN	OPOL	>90%	79	Poor
28.	JUN	F	10	ENG	JPN	OPOL	>90%	68	Very Poor

Table 2. TOWL results and language use in the family.

	Child	Sex	Age	NJP’s Language Use	JP’s Language Use	Current Home Exposure	Current Child ENG Use	Score	Descriptive Terms
1.	LIS	F	12	ENG	JPN	OPOL	>90%	124	Superior
2.	SAK	F	10	ENG-JPN	JPN	More ENG	>90%	123	Superior
3.	YOD	M	14	ENG	JPN	OPOL	>90%	123	Superior
4.	LUH	M	9	ENG	JPN	OPOL	>90%	119	Above Average
5.	AYO	F	12	ENG	JPN	OPOL	>90%	118	Above Average
6.	HUG	M	9	ENG	JPN	OPOL	>90%	115	Above Average
7.	EMS	F	9	ENG	JPN	OPOL	>90%	109	Average
8.	KAT	M	9	ENG	ENG-JPN	More ENG	>90%	109	Average
9.	KOO	F	9	ENG	JPN-ENG	More ENG	>90%	106	Average
10.	TAK	M	9	ENG	JPN	OPOL	>90%	104	Average
11.	MAS	F	11	ENG	ENG-JPN	More ENG	>90%	103	Average
12.	AMS	F	11	ENG	JPN-ENG	More ENG	>90%	102	Average
13.	KAN	F	10	ENG-JPN	JPN-ENG	More ENG	75–90%	102	Average
14.	LUM	M	9	ENG-JPN	JPN-ENG	More ENG	60–75%	101	Average
15.	JAS	M	10	ENG	JPN	OPOL	>90%	100	Average
16.	ERS	F	10	ENG	JPN-ENG	More ENG	>90%	96	Average
17.	MOH	M	11	ENG	ENG-JPN	More ENG	>90%	95	Average
18.	ARS	F	12	ENG	JPN-ENG	More ENG	30–45%	95	Average
19.	YUK	M	9	ENG-JPN	JPN-ENG	More ENG	75–90%	95	Average
20.	TYA	M	12	ENG	JPN-ENG	More ENG	>90%	94	Average
21.	LET	M	9	ENG	JPN	OPOL	>90%	94	Average
22.	ISN	M	12	ENG	JPN	OPOL	60–75%	92	Average
23.	KYM	F	12	ENG-JPN	JPN-ENG	More ENG	75–90%	91	Average
24.	CAS	M	14	ENG	JPN	OPOL	>90%	90	Average
25.	ALN	M	12	ENG	JPN	OPOL	>90%	86	Below Average
26.	SHS	M	10	ENG-JPN	JPN-ENG	OPOL	75–90%	83	Below Average
27.	JOG	M	9	ENG	JPN	OPOL	>90%	79	Poor
28.	JUN	F	10	ENG	JPN	OPOL	>90%	68	Very Poor

3.2. Spontaneous Writing versus Contrived Writing

Table 3. Contrived writing and spontaneous writing scores (N = 28).

	Contrived Writing		Spontaneous Writing		Overall Writing		Group
	No.	(%)	No.	(%)	No.	(%)
Very Superior (>130)	0	(0.0%)	3	(10.7%)	0	(0.0%)	1
Superior (121–130)	0	(0.0%)	10	(35.7%)	3	(10.7%)	1
Above Average (111–120)	5	(17.9%)	7	(25.0%)	3	(10.7%)	1
Average (90–110)	11	(39.3%)	6	(21.4%)	18	(64.3%)	2
Below Average (80–89)	10	(35.7%)	0	(0.0%)	2	(7.1%)	3
Poor (70–79)	1	(3.6%)	2	(7.1%)	1	(3.6%)	3
Very Poor (<70)	1	(3.6%)	0	(0.0%)	1	(3.6%)	3

shows that more children performed better in the spontaneous writing section than in the contrived writing section of the TOWL. Twenty children had scores ranging from Above Average to Very Superior for spontaneous writing, but only five were in the Above Average range for contrived writing. Twelve children scored within the lower score ranges (Very Poor to Below Average) for contrived writing, but only two children were in this low range for spontaneous writing. These two were our younger participants (aged 9 and 10), who were inexperienced with both the story composition and contextual convention aspects of the spontaneous writing section. Many children were weak in the vocabulary and spelling subtests of contrived writing but performed well in their story composition in spontaneous writing, where they adeptly created storylines and assigned emotions to their story’s characters. Their overall writing reflects their different results in the two sections of the TOWL and gives us three groups of performers.

A writing sample can illustrate spontaneous writing abilities clearly. MAS’s overall writing score of 103 falls under Average (Table 2, no. 11). However, her spontaneous writing score of 121 (Superior) was much higher than her contrived writing score of 96 (Average). Figure 1 is an excerpt from the first half of MAS’s story, based on a stimulus picture of a burning tree in a thunderstorm. The story had many spelling mistakes (e.g., whin “when”, clovs “clothes”, and socing “soaking”), and a noun–verb disagreement in the first sentence. However, she produced a well-sequenced story that went beyond the picture. She composed several compound sentences and assigned dialogue to her characters using quotation marks, question marks, and apostrophes. To provide an additional perspective on her writing performance, we can examine her story using the Common European Framework of Reference for Languages (CEFR 2020) for “creative writing”. MAS’s story is close to Level B2, an upper intermediate level, because she could provide “clear, detailed descriptions of real or imaginary events and experiences marking the relationship between ideas in clear connected text, and following established conventions of the genre concerned” (p. 67). Other children also performed well by producing creative and well-sequenced stories in 20 min. Most children’s writing was at least at the B1 level, the fourth highest of six CEFR levels, because they could “narrate a story”, “describe an event”, and “give accounts of experiences, describing feelings and reactions in simple, connected text”. These descriptors further illustrate our participants’ developing writing ability in English.

3.3. Home and Weekend School Factors

Table 4 presents the mean scores from 22 questionnaire items that parents rated on seven-point scales for agreement, frequency, or duration. Six statements each (not in consecutive number order) are grouped according to whether they pertain to home reading practices, home writing practices, or parental support, respectively. Three statements and an open response about the length of attendance at weekend school relate to weekend school factors. Please refer to the Appendix B of the questionnaire, for the complete statements corresponding to the number and their abbreviated descriptions shown in Table 4. The scales corresponding to the item numbers are as follows:

• Nos. 5–11 and 16 are based on an agreement scale: (1) Don’t know, (2) Strongly Disagree, (3) Disagree, (4) Somewhat Disagree, (5) Somewhat Agree, (6) Agree, and (7) Strongly Agree.
• Nos. 19–22, 26–28, 31–32, and 34–37 are based on a frequency scale: (1) Never, (2) Occasionally, (3) Infrequently, (4) Somewhat Infrequently, (5) Somewhat Frequently, (6) Frequently, and (7) Very Frequently.
• Length of attendance (years) is based on mapping time at weekend school to seven categories: (1) < one year, (2) 1–2 years, (3) 2–3 years, (4) 3–4 years, (5) 4–5 years, (6) 5–6 years, and (7) > 6 years.

Table 4. Mean scores for questionnaire items related to home and weekend school factors with Spearman’s correlation results between writing scores and those factors (N = 28).

Home reading	5	6	19	26	27	28
	Enjoy reading	Prefer reading	Reading frequency	Independent reading	Voluntary reading	Choose books
Mean	4.8	3.8	5.5	4.5	3.7	4.8
SD	1.6	1.9	1.1	2.1	2.1	2.1
Correlation Coeff.	0.665 **	0.554 **	0.318	0.631 **	0.638 **	0.619 **
Sig. (2-tailed)	0.000	0.002	0.099	0.000	0.000	0.000
Home writing	7	8	20	31	35	36
	Enjoy writing	Prefer writing	Writing frequency	Communicative writing	Additional writing	Writing feedback
Mean	3.8	3.6	5.1	3.2	3.1	4.9
SD	1.4	1.5	1.1	1.6	1.7	1.5
Correlation Coeff.	0.553 **	0.372	0.063	0.393 *	−0.194	−0.151
Sig. (2-tailed)	0.002	0.051	0.749	0.039	0.322	0.443
Parental support	9	10	11	21	34	37
	Plenty books	Appropriate books	Literacy time/place	Parental involvement	Give rewards	Make corrections
Mean	6.5	6.2	4.4	5.5	2.8	5.4
SD	0.6	0.8	1.5	0.9	1.8	1.2
Correlation Coeff.	0.112	0.339	−0.043	−0.032	−0.449 *	−0.292
Sig. (2-tailed)	0.569	0.077	0.829	0.873	0.017	0.132
Weekend school	16	22	32	Length of attendance
	Enjoy school	Attend school	Complete homework	(in years for M, in months for rs)
Mean	5.8	6.4	5.8	5.4
SD	1.3	0.9	1.3	1.8
Correlation Coeff.	0.383 *	0.262	0.180	0.062
Sig. (2-tailed)	0.044	0.178	0.361	0.753

** Correlation is significant at the 0.01 level (two-tailed). * Correlation is significant at the 0.05 level (two-tailed).

Table 4. Mean scores for questionnaire items related to home and weekend school factors with Spearman’s correlation results between writing scores and those factors (N = 28).

Home reading	5	6	19	26	27	28
	Enjoy reading	Prefer reading	Reading frequency	Independent reading	Voluntary reading	Choose books
Mean	4.8	3.8	5.5	4.5	3.7	4.8
SD	1.6	1.9	1.1	2.1	2.1	2.1
Correlation Coeff.	0.665 **	0.554 **	0.318	0.631 **	0.638 **	0.619 **
Sig. (2-tailed)	0.000	0.002	0.099	0.000	0.000	0.000
Home writing	7	8	20	31	35	36
	Enjoy writing	Prefer writing	Writing frequency	Communicative writing	Additional writing	Writing feedback
Mean	3.8	3.6	5.1	3.2	3.1	4.9
SD	1.4	1.5	1.1	1.6	1.7	1.5
Correlation Coeff.	0.553 **	0.372	0.063	0.393 *	−0.194	−0.151
Sig. (2-tailed)	0.002	0.051	0.749	0.039	0.322	0.443
Parental support	9	10	11	21	34	37
	Plenty books	Appropriate books	Literacy time/place	Parental involvement	Give rewards	Make corrections
Mean	6.5	6.2	4.4	5.5	2.8	5.4
SD	0.6	0.8	1.5	0.9	1.8	1.2
Correlation Coeff.	0.112	0.339	−0.043	−0.032	−0.449 *	−0.292
Sig. (2-tailed)	0.569	0.077	0.829	0.873	0.017	0.132
Weekend school	16	22	32	Length of attendance
	Enjoy school	Attend school	Complete homework	(in years for M, in months for rs)
Mean	5.8	6.4	5.8	5.4
SD	1.3	0.9	1.3	1.8
Correlation Coeff.	0.383 *	0.262	0.180	0.062
Sig. (2-tailed)	0.044	0.178	0.361	0.753

** Correlation is significant at the 0.01 level (two-tailed). * Correlation is significant at the 0.05 level (two-tailed).

Overall, a few factors were common among the families, as reflected in mean scores higher than six for the parental support and weekend school factors. Many parents “agree” that their children have plenty of English reading materials (M = 6.5, SD = 0.6) that are age- and genre-appropriate (M = 6.2, SD = 0.8). Many of their children attend weekend school “frequently” (M = 6.4, SD = 0.9), i.e., 75% to 90% of the time. To a lesser extent, the parents also “agree” that their children enjoy weekend school (M = 5.8, SD = 1.3) and “frequently” complete their weekend school homework (M = 5.8, SD = 1.3). These results indicate that many children’s learning is supported by having adequate and suitable books at home, regularly attending and enjoying weekend school, and by doing homework.

To understand better the relationship between the children’s writing scores and the four categories of literacy-related factors, Spearman’s rank correlation was computed as a supplementary analysis and also summarized in Table 4. Out of 22 literacy-related factors, a strong relationship can be found mainly between the writing scores and home reading practices. There was a statistically significant strong positive correlation between the scores and five out of six variables: no. 5 enjoys reading in English (r = 0.67, p < 0.001), no. 6 prefers to read in English than in Japanese (r = 0.55, p = 0.002), no. 26 reads independently in English (r = 0.63, p < 0.001), no. 27 reads voluntarily in English (r = 0.64, p < 0.001), and no. 28 chooses own English reading materials (r = 0.62, p < 0.001). The relationships between the scores and factors in the other three categories are weaker and fewer in number. In “Home writing practices”, there was a statistically significant strong positive correlation for no. 7, enjoys writing in English (r = 0.55, p = 0.002) and a moderately significant correlation for no. 31, writes in English for communicative purposes (e.g., diaries, letters) (r = 0.39, p = 0.039). There was a moderately significant negative correlation between the scores and one item in “Parental support”, no. 34, receives rewards for doing reading and writing tasks (r = −0.45, p = 0.017). There was also a moderately significant positive correlation between the scores and one item in “Weekend school factors”, no. 16, enjoys weekend English school (r = 0.38, p = 0.044).

To determine the home and weekend school factors that contribute to different levels of writing, we grouped the children based on their writing scores; as mentioned previously, Group 1 (n = 6) scored above 111 (Above Average or higher), Group 2 (n = 18) scored between 90 and 110 (Average), and Group 3 (n = 4) scored less than 90 (Below Average or lower). We then analyzed the specific factors associated with each group. Parental responses to questionnaire items related to these factors are summarized in box plots in Figure 2, Figure 3, Figure 4 and Figure 5. The boxes represent the range between the 25th and 75th quartiles, and the upper and lower lines show the minimum and maximum responses given. The cross within the box indicates the mean score, whereas the horizontal line represents the median. Outliers are shown by small circles.

Figure 2 summarizes parental responses to home reading practices. It shows that Group 1 (unlike Groups 2 and 3) has strong reading practices, with mean scores of 5.5 or higher for all six reading variables. The parents tended to “agree” that their children enjoy reading in English (M = 6.3, SD = 0.8), prefer to read in English than in Japanese (M = 5.5, SD = 1.6), and choose their own books (M = 6.3, SD = 1.0). All three groups read in English, but in decreasing frequency from “frequently” for Group 1 (M = 6.0, SD = 0.8), “somewhat frequently” for Group 2 (M = 5.4, SD = 1.1), and slightly less than “somewhat frequently” for Group 3 (M = 4.8, SD = 1.0). The practices of reading independently (M = 6.5, SD = 0.8) and reading voluntarily (M = 5.7, SD = 1.2) are more frequent for Group 1 than Groups 2 and 3, indicating that Group 1 children have established a habit of reading for pleasure. The parents of Group 3 children tended to “disagree” that their children read independently and voluntarily; thus, more parental effort would be required to encourage children to do so.

Figure 2. Parental questionnaire responses related to home English reading practices.

Figure 2. Parental questionnaire responses related to home English reading practices.

The box plot in Figure 3 shows responses related to home writing practices. The mean scores for writing practices are generally lower than for reading practices, indicating that writing is less practiced in the home. Even Group 1 children were not particularly fond of writing; their parents only “somewhat agree” or even “somewhat disagree” that they enjoy writing (M = 4.8, SD = 1.5) and prefer writing in English than in Japanese (M = 4.5, SD = 1.2). However, these mean scores are even lower for children in Groups 2 and 3. Particularly, Group 3 parents tended to “strongly disagree” or “disagree” that their children enjoy writing in English (M = 2.8, SD = 0.5) or that they prefer to write in English than in Japanese (M = 2.3, SD = 0.5). No great differences can be seen for the four remaining writing variables. All three groups only write and receive feedback “somewhat frequently”, with lower frequencies for writing communicatively in English and doing additional writing at home.

Figure 3. Parental questionnaire responses related to home English writing practices.

Figure 3. Parental questionnaire responses related to home English writing practices.

Likewise, Figure 4 shows little difference in parental support and the home literacy environment. All three groups have plenty of genre- and level-appropriate reading materials, and their parents put in a similar amount of time and effort in home literacy activities. The Spearman statistical test revealed a moderately significant negative correlation between the scores and giving rewards for literacy activities. Interestingly, parents tended to reward Group 3 children more frequently, i.e., 2−3 times/month (M = 4.8, SD = 1.0), than Group 2 children (M = 2.7, SD = 1.8), who may be rewarded every few months for doing English literacy tasks. Meanwhile, five out of six Group 1 children never received any rewards (M = 1.5, SD = 1.1). Parents of Group 3 children tended to correct their reading and writing more often, 3−5 days/week (M = 6.0, SD = 1.4), while corrections were given once/week for Group 2 (M = 5.4, SD = 1.3) and Group 1 (M = 5.2, SD = 0.4) children.

Figure 4. Parental questionnaire responses related to parental support and home literacy environment.

Figure 4. Parental questionnaire responses related to parental support and home literacy environment.

Figure 5 shows similarities between the three groups for factors related to weekend school. Most of the children attended weekend school “frequently” and completed their homework “frequently” or “somewhat frequently”. While Group 3 children may not enjoy weekend school as much as the other two groups, most of the children have been attending for four years or more. In the next section, we will discuss the outliers shown in Figure 2, Figure 3, Figure 4 and Figure 5.

Figure 5. Parental questionnaire responses related to weekend school factors.

Figure 5. Parental questionnaire responses related to weekend school factors.

4. Discussion

4.1. The Importance of Weekend Schools for English Literacy in Japan

Regarding our first research question on how well Japanese-English bicultural children attending English weekend schools can write in English, 24 of our 28 participants performed at average and above levels as same-aged US children. While most of the children in Group 1 attended weekend school very frequently (>90% of the time), the outlier, YOD (Table 2, no. 3), attended somewhat infrequently (45–60% of the time) because he was already a middle school student who was busy preparing for high school entrance examinations. His superior writing performance can be attributed not only to six years of weekend school attendance but the fact that his American mother is an English teacher who often supports his English learning. Almost half of our non-Japanese parents are language teachers, corroborating Fu’s (2020) suggestion that parents who are teachers facilitate children’s home language development.

We then examined the home and weekend school factors of stronger and weaker writers. While there were no significant associations between the writing scores and the length of enrollment, notably, the top performer, LIS (Table 2, no. 1), attended weekend school the longest of all the children at ten years, starting from age 2. The second top performer, SAK, however, has only attended for 2.5 years, one of the shortest lengths of enrollment among the children. SAK, however, attended kindergarten in the US, while interestingly, so did LIS, albeit only for less than two months. Note that LET, the only child born in the US who lived there until age six, did not have an advantage, as his score of 94 is on the lower end of Group 2 (Table 2, no. 21). All Group 3 children with the lowest scores have attended weekend school longer than SAK, with two having attended for six years. This finding reveals a lack of association between writing scores and length of enrollment, confirming Kim and Pyun’s (2014) observation that literacy skills do not improve with years of weekend school attendance. This result also points to the importance of parental support and the home literacy environment.

4.2. Parental Support and the Home Literacy Environment

Interestingly, SAK, with the second highest score but the shortest time at weekend school, is the only child from Group 1 who received any rewards (e.g., game time) for doing reading and writing tasks, albeit somewhat infrequently (45–60% of the time). In sharp contrast to the lack of rewards for most of the Group 1 children and the occasional rewards (15–30% of the time) for Group 2 children, Group 3 children were rewarded more often. However, these incentives did not translate into their writing performance (Figure 4). Thus, extrinsic motivation did not seem to work. For all three groups, parents were frequently (60–90% of the time) expending time and effort to help their children and correcting their work. Not surprisingly, Group 3 children received help the most frequently. Only two children in Group 2 did not receive much feedback from their parents. All our participants had adequate and suitable English reading materials at home. Furthermore, the Tokyo weekend school is located near a bookstore that is well-stocked with children’s English books, and parents also conducted book swaps to give their children a greater variety of reading materials.

4.3. Reading and Writing Practices at Home

As noted above, differences in writing performance were not associated with parental support, home literacy environment, and weekend school factors. Interestingly, reading rather than writing practices at home had a stronger relationship with the children’s writing; those with higher writing scores enjoyed and preferred to read in English and did so voluntarily and independently. A comparison of mean scores from parental questionnaire responses revealed that writing practices did not differ much between our three groupings. Group 1 children did not write, do additional writing homework, or receive feedback from their parents more often than the other two groups. Home writing practices were less frequent than reading practices, probably because it is not easy for parents to teach writing. The TOWL test creators, Hammill and Larsen (2009), suggest that a high performance indicates not only a command of a wide range of writing skills, but also a good reading ability. Our study found that the Group 1 high performers read more in English, and they often read independently and voluntarily. They also preferred to read in English than in Japanese more often than the other two groups. Regular reading inadvertently improved not only their reading fluency but also their writing. The interesting storylines, coherent sequences, and sustained pace of Group 1′s stories in the TOWL’s spontaneous writing section reflected their exposure to many storybooks in English. Group 1 children rarely received rewards from their parents for reading and writing in English.

As in our study, Guthrie et al. (2007) and Schiefele et al. (2012) note that intrinsic motivation and perceived autonomy can improve reading comprehension and increase the amount of reading in the long term. Promoting the habit of reading English independently for pleasure may be a useful way for parents to improve their children’s English writing than doing reading or writing together with them. These findings confirm the results of studies showing the benefit of recreational reading for heritage language literacy learning (Cho and Krashen 2000; Taniguchi 2021; Tse 2001). While interest in recreational English reading may be a child internal factor that is difficult to control, parents can instill good reading practices by modeling reading and introducing books that spark children’s interest. Weekend schools can also play a role by creating a mini library, organizing book swaps, and encouraging book discussions. Recreational reading may help maintain biliteracy when teenagers no longer want to participate in more structured literacy activities with their parents and in weekend school.

Fostering children’s motivation to read for pleasure, e.g., through print exposure and shared book reading, would benefit literacy development in the home language at older ages. When children become proficient readers, they are also likely to write better. Reading and writing are closely related because they draw from some common cognitive processes and knowledge bases: metaknowledge (i.e., self-monitoring of one’s knowledge), content knowledge about the world, knowledge about text attributes, and procedural knowledge (i.e., creating and using meaningful, connected texts) (Fitzgerald and Shanahan 2000). Ahmed et al. (2014) found that the effect of reading on writing is particularly strong: word recognition predicts word spelling, sentence reading predicts sentence construction, and reading fluency predicts writing fluency. Our results thus suggest that weekend school is essential for literacy learning, but it is the home reading practices that contribute to stronger writing skills.

5. Conclusions

Formal language assessments are hardly conducted in weekend schools. The writing assessment results in this study show that many of our child participants possessed age-appropriate English writing ability; their writing ability was at the Average level, with some performing at the Above Average level or higher. Their performance demonstrates simultaneous Japanese-English bilingual and bicultural children’s potential for mastering literacy in both languages, despite attending regular Japanese-medium schools in Japan and receiving only limited English literacy instruction in weekend school. This finding shows that intensive exposure through studying abroad and attending bilingual schools (Taura and Taura 2012) are not the only ways for Japanese-English bilingual children to develop their English literacy. Weekend schools can be an affordable and accessible model for successful home language literacy learning, even in late childhood and adolescence. However, analyses of home and weekend school factors affecting individual writing performance revealed that, while weekend school played a fundamental role in English literacy learning, home reading practices contributed to better writing ability. The strongest writers in our study were the most eager readers; they enjoyed reading for pleasure. Weaker writers, however, continued to participate in parent-initiated joint reading. This result underscores the importance of fostering independent recreational reading to promote strong home language literacy skills in older children.

While we determined specific home and weekend school factors affecting English writing, we acknowledge the influence of broader sociolinguistic factors, namely the prestige of English in Japan. Good English skills are necessary for entering prestigious schools or studying abroad, both of which are common expectations of parents of children who attend weekend schools (Nakamura 2019). While English-speaking parents may hold positive attitudes and strong impact beliefs about raising bilingual and biliterate children, speakers of other minority languages may feel differently, making it challenging to develop literacy skills or even oral ability in the home language (Kitayama 2012; Nakamura 2016; Saito 2005). Our study is limited by its small sample, which may be skewed toward more confident writers from families that are highly motivated to develop English literacy. To understand better Japanese-English bilingual and bicultural children’s long-term English literacy development, we will include reading assessments, and monitor reading and writing development longitudinally for a larger group of children in the next phase of our research.