Introduction

China is a multiethnic country. There are 56 ethnic groups, of whom 55 are the minority groups. The policy around reform and openness in China has led schools and institutions to strive to improve minority students’ English language proficiency. However, provision of English learning to minority students has been criticized as insufficient due to an acute shortage of bilingual teachers and materials and the lack of a sociolinguistic environment (Feng & Adamson, 2014; Hu, 2007). In addition, ethnic minority students are often at a disadvantage within the education system; they encounter challenges in obtaining equal opportunities and resources (Feng & Adamson, 2018). Despite extensive research on multilingual and multicultural development, little international scholarly attention has focused on ethnic minority populations in China.

Metacognitive knowledge and vocabulary knowledge are commonly acknowledged as two important variables for learners. The role of metacognitive knowledge in listening comprehension (Goh & Vandergrift, 2022; Vafaee & Suzuki, 2020; Zhang & Goh, 2006) and reading comprehension (Brown, 1987; van Steensel et al., 2016; Zhang, 2010) has received considerable attention. In recent years, researchers began to pay attention to the role of metacognitive knowledge in learners’ acquisition of vocabulary knowledge (e.g., Teng, 2022). One of the reasons behind this effort may be related to the fact that the more a learner is willing to be engaged with new words, the more likely he/she is to learn them (Schmitt, 2010). Indeed, the more attention given to an item, and the more manipulation involved with it, the greater the chances it will be learned and remembered (Teng & Zhang, D., 2021). However, learners in a foreign language context, particularly those ethnic minority learners, may lack metacognitive awareness to take control of their vocabulary learning. In Lepola et al.’s (2020) study, vocabulary, metacognitive knowledge and task orientation were evaluated at the age of 5. Latent growth curve modeling showed a pattern of decreasing achievement gaps in narrative picture book comprehension. Vocabulary and metacognitive knowledge uniquely contributed to the concurrent level of narrative picture book comprehension. The empirical evidence from studies such as Lepola et al (2020) suggests concurrent association between metacognitive knowledge and vocabulary at the age of 5. Their findings also suggested unique contributions of vocabulary, metacognitive knowledge and task orientation to the initial level as well as to the growth of comprehension from age 5 to 9. Further empirical evidence suggests longitudinal and dynamic development between metacognitive knowledge and vocabulary knowledge, and that metacognitive knowledge can influence the development of vocabulary knowledge (Teng, 2022). Early research (e.g., Annevirta & Vauras, 2001) also shows that primary school learners’ metacognitive knowledge can increase along with their age and learning experience. However, these findings did not delineate the development of metacognitive knowledge in ethnic minority students’ vocabulary knowledge, let alone comparing minority and non-minority students’ development of metacognitive knowledge and vocabulary knowledge.

The current study aims to bridge this gap through investigating a sample of ethnic minority (Yao) and ethnolinguistic non-minority students (Han), with a focus on their early development of metacognitive knowledge and breadth of vocabulary knowledge from third through sixth grade. The minority group, the Yao, spoke a language at home that differed from the language of instruction in the schools they attended. The majority group, the Han, spoke Mandarin at home and at school. The Yao children therefore needed to learn to speak Mandarin at school, which was their L2. The different language status may influence their English vocabulary learning. The present study attempted to assess the possible mutual relationship between metacognitive knowledge and breadth of English vocabulary knowledge among Han and Yao students. It intends to shed light on the different development trends of metacognitive knowledge and breadth of English vocabulary knowledge in Han and Yao students. The findings can contribute to an understanding of ethnolinguistic minority learners’ characteristics in developing metacognitive knowledge and breadth of English vocabulary knowledge.

Literature review

Metacognitive knowledge

Flavell (1979) described metacognitive knowledge as “knowledge or beliefs about what factors or variables act and interact in what ways to affect the course and outcome of cognitive enterprises” (p. 907). It includes (a) learning processes and beliefs about learning, (b) the task of learning and processing information, and (c) strategies for learning. Metacognitive knowledge functions as a prerequisite for a learner to regulate his or her cognitive activity. Brown (1987) described metacognitive knowledge as including three aspects: declarative knowledge (knowledge of what one knows and how to learn as well as the factors that influence the learning process), procedural knowledge (knowledge about how to perform a specific skill or task to achieve specific learning goals), and conditional knowledge (knowledge of external conditions and an awareness of certain strategies that can be applied to those conditions). Schraw (1994) described metacognitive knowledge as people’s knowledge about cognition in general, i.e., person knowledge (i.e., learners’ knowledge of their own cognitive abilities), task knowledge (i.e., knowledge of task features and processing demands), and strategy knowledge (i.e., cognitive and metacognitive strategies for solving problems or completing tasks). Metacognitive knowledge, as described by Efklides (2006, 2014), encompasses our declarative understanding of persons, tasks, strategies, and goals. It encompasses the knowledge retrieved from memory, the beliefs and theories regarding cognition and its operation, the understanding of knowledge and its validity criteria, as well as the insights into one’s own thinking processes and the thinking of others. According to Efklides (2008, 2011), individuals must possess a conscious understanding of their objectives and effectively monitor and regulate their cognition, emotions, behavior, and environment in order to successfully engage in self-regulation. Metacognition, including metacognitive knowledge, plays a vital role in facilitating this process.

In particular, an awareness of person, tasks, and strategies reflects learners’ abilities to remember, understand, and learn new knowledge (Annevirta et al., 2007). An awareness of person, tasks, and strategies was the focus of the instrument on metacognitive knowledge in the present study. In a recent study, Marulis et al. (2020) found that kindergarten and primary students had already developed metacognitive knowledge, and a higher level of metacognitive knowledge facilitated these young learners’ processing of new information, execution of new tasks, reflection on their strengths and weaknesses, and exertion of efforts for their own learning. Teng and Zhang, L., (2021) also suggested the possibility for learners to develop metacognitive knowledge during the primary school grades, and the development of metacognitive knowledge predicts learners’ reading and writing proficiency. Again, Teng (2022) suggested the role of metacognitive knowledge in vocabulary knowledge. This longitudinal study included a sample of 426 first-grade students. A series of tests measuring metacognitive knowledge and vocabulary knowledge were administered four times. The metacognitive knowledge test involved one-on-one assessments where students provided explanations regarding cognitive activities. The vocabulary knowledge test assessed students’ breadth and depth of vocabulary knowledge. Latent growth curve modeling was employed to examine developmental changes over time. The findings revealed that participants’ metacognitive knowledge and vocabulary knowledge displayed improvement from 1st to 4th grade, although the growth patterns were not strictly cumulative. Notably, participants’ level of metacognitive knowledge exhibited a strong association with their vocabulary knowledge throughout the assessed school years. It appears that young learners’ metacognitive knowledge can be assessed at the primary school level, and early metacognitive knowledge is a potent predictor of language learning, especially the development of vocabulary knowledge.

Vocabulary knowledge

Vocabulary knowledge is an essential foundation for second and foreign language acquisition (Qian, 2002; Schmitt, 2014). In the case of young learners, both vocabulary and syntax are crucial for their literacy development and development of later communication skills throughout the whole school years (Harris et al., 2011). According to Schmitt (2014), breadth of vocabulary knowledge refers to the number of words the meaning of which one has at least some superficial knowledge, e.g., meaning recognition assessed in the present study. Depth of vocabulary knowledge refers to how well a learner knows about a word (Schmitt, 2014), including knowledge of spoken and written form, morphological knowledge, word meaning knowledge, collocational and grammatical knowledge, connotative and associative knowledge, and the knowledge of social or other constraints in using a word (Laufer & Goldstein, 2004). Learners should develop rich knowledge of each individual lexical item to help them function within a language. Knowing a word involves possession of a variety of different aspects of knowledge, and learners may have knowledge of these aspects at different levels of strength, detail, and fluency (Nation, 2020).

The acquisition of vocabulary knowledge may be dependent on various factors, including contextual clues (Webb, 2007), word exposure frequency (Teng, 2019, 2020; Webb, 2007), sentence context (Chilton & Ehri, 2015), vocabulary learning strategies (Gu & Johnson, 2016), and metacognitive knowledge for self-regulated learning (Teng & Zhang, D., 2021). In an early survey study, Gu and Johnson (1996) found that vocabulary knowledge was dependent on metacognitive strategies such as self-initiation and selective attention. Vocabulary learning requires a multiplicity of cognitive processes during which learners can obtain vocabulary knowledge as they grow, as shown in a longitudinal study (Sparks & Deacon, 2015). Acquiring vocabulary knowledge is thus a cumulative process in terms of both increasing the number of words known and increasing the depth of knowledge of words (Read, 2004). Vocabulary knowledge grows and becomes strengthened over time rather than being a process of full, immediate acquisition. According to Paris (2005), vocabulary knowledge is an unconstrained skill, and vocabulary knowledge is supported by and supports linguistic, cognitive, and communicative skill development. The two major conditions in vocabulary learning are the number of meetings and quality of attention (Nation, 2020). Essentially, learners may find some words very easy to learn, while other words require higher degrees of effort.

Several studies have highlighted the significant impact of vocabulary knowledge on students’ reading proficiency (Qian, 2002). In comparison to average comprehenders, both poor Chinese and poor English comprehenders demonstrated a lack of proficiency in Chinese lexical tone awareness. However, there were no notable differences in scores related to segmental phonological awareness and vocabulary knowledge, regardless of whether it was assessed in Chinese or English (Choi et al., 2017). Additionally, research indicates that learners from diverse backgrounds, such as children from economically disadvantaged households, tend to acquire a smaller vocabulary compared to their peers from more affluent homes (Hart & Risley, 1995). Nevertheless, there is a notable absence of research investigating the individual differences among primary school students when it comes to acquiring breadth of vocabulary knowledge, particularly in the context of comparing minority and non-minority students. The primary impetus for this study stems from the dearth of research findings.

Longitudinal studies on metacognitive knowledge and vocabulary knowledge

Examining longitudinal development of metacognitive knowledge and vocabulary knowledge is meaningful. In terms of metacognitive knowledge, Flavell (1979) first proposed a model of cognitive monitoring. This model sheds light on the development of metacognitive knowledge for young learners. For example, children first attempt to distinguish between understanding things and not understanding things. They then attempt to process puzzling, confusing, and uncertain input and finally proceed toward a clear representation of something and a definite sense of what they should do next. Longitudinal studies have also been conducted to delineate children’s development of metacognitive knowledge. For example, Annevirta and Vauras (2001) explored 196 primary school students’ longitudinal development of metacognitive knowledge from preschool to third grade. The results showed that it is possible for young learners to develop significantly and cumulatively in terms of their metacognitive knowledge during the first three years. While the observed phenomenon does not directly parallel the Matthew effect, which entails the accumulation of advantages for the already advantaged and the struggle to catch up for the disadvantaged, the findings affirm that young learners demonstrated notable growth in their comprehension and mindfulness of their own cognitive processes, learning tactics, and abilities as time progressed. In another longitudinal study focusing on primary school students’ development of metacognitive knowledge (Annevirta et al., 2007), the results showed individual differences in the development of metacognitive knowledge. The greater the positive change in children’s metacognitive knowledge from preschool to the 2nd grade, the higher their text comprehension abilities demonstrated during the first three years of school. Latent growth curve (LGC) modeling showed that no uni-construct effect was detected for metacognitive knowledge and text comprehension. However, slight multi-construct cumulative development was found between the children’s metacognitive knowledge and text comprehension. The uni-construct effect pertains to a scenario in which individuals who already have certain advantages or resources tend to amass additional advantages. Multi-construct cumulative development refers to a process in which multiple factors or constructs contribute to the gradual and cumulative development of an individual or a system. The findings thus suggest that the development of metacognitive knowledge is a complex and multifaceted phenomenon. In a recent study on 420 Chinese students’ longitudinal development of metacognitive knowledge, reading and writing performance (Teng & Zhang, L., 2021), the results showed that learners who possessed a higher level of metacognitive knowledge were better able to develop a higher level of metacognitive knowledge in their subsequent years of learning. The findings from above studies support the longitudinal development of metacognitive knowledge.

In another longitudinal study that involved 300 university students, Zhang and Lu (2014) collected data from a vocabulary level test at three points in time over the course of 22 months and showed the effect of frequency level on acquiring breadth of vocabulary knowledge and vocabulary fluency. Breadth of vocabulary knowledge is a more difficult dimension to acquire than vocabulary fluency. In a recent study, Teng (2022) used a growth curve modeling approach to examine primary school learners’ longitudinal development of metacognitive knowledge and vocabulary knowledge. The results supported individual differences in the development of the two types of knowledge. The level of metacognitive knowledge determines the development of vocabulary knowledge. Teng and Zhang (2022) conducted a longitudinal study that specifically examined the metacognitive knowledge and morphological awareness of nonminority Han students and minority Yao students. The study findings provided support for the interconnected development of metacognitive knowledge and morphological awareness, with a particular emphasis on the role of metacognitive knowledge on morphological awareness. However, the results indicated that Yao students significantly fell behind Han students, despite both groups displaying a cumulative increase in metacognitive knowledge and morphological awareness from the third to the sixth grade. Overall, previous studies shed light on the relationship between metacognitive knowledge and vocabulary knowledge, but relatively little is known about the longitudinal development of metacognitive knowledge and vocabulary knowledge, particularly in ethnic minority students.

There were some studies that might focus on minority students. For example, in a longitudinal study of minority students in the United States by Mancilla-Martinez and Lesaux (2010), the focus was on the growth in English and Spanish word reading and vocabulary and how the growth rates of word reading and vocabulary would predict English reading outcomes. Students were followed from ages 4.5 to 11 years. The results supported the proposition that learners’ outcomes in English reading are dependent on the initial level and growth of reading and vocabulary knowledge. In another longitudinal study (Grimm et al., 2018), latent growth curve modeling was adopted to investigate 193 Spanish-speaking minority students’ English development from third to eighth grade. The results showed that minority students scored lower than the average level in English literacy in California. In addition, their literacy development plateaued after fifth grade. A variety of factors, including letter knowledge, phonological awareness, word reading, and vocabulary, predicted English development. The findings provide insight into the learning difficulties encountered by minority students.

We may expect growth in vocabulary knowledge in a group of minority students, although this growth would be less than for non-minority students. However, the data comparing minority and nonminority students have not been sufficient. Minority students may develop limited vocabulary knowledge because of a prolonged lack of language learning resources and language input (e.g., Shahar-Yames & Prior, 2018). However, whether differences in growth rates exist between minority and non-minority students, particularly in the Chinese context, remains an open question.

The present study

The present study aims to examine the growth of metacognitive knowledge and breadth of vocabulary knowledge in two groups of minority (Yao) and non-minority (Han) students in China. In particular, we examine whether there are differences between minority and non-minority students in their development of metacognitive knowledge and breadth of vocabulary knowledge. This study allows us to examine breadth of vocabulary knowledge as well as how such vocabulary knowledge is influenced by metacognitive knowledge from grades 3 to 6. It aims to address three research questions:

  1. 1.

    Do minority students and non-minority students demonstrate differences in their development of metacognitive knowledge and breadth of vocabulary knowledge?

  2. 2.

    How do Han and Yao students develop metacognitive knowledge and breadth of vocabulary knowledge from grades 3 to 6?

  3. 3.

    How is the level and development of metacognitive knowledge from grades 3 to 6 related to the level and development of breadth of vocabulary knowledge?

Method

Participants

The longitudinal data were from a sample of 223 primary school third-grade students. Among the students, 115 learners (male: 67; female: 48) were from the Yao ethnic minority group and 108 learners (male: 52; female: 56) were from the non-minority group, also referred to as Han. In the present study, Han people who speak Mandarin before primary school are defined as nonminority group members. The Yao people (also known as Mien) are a government classification as an ethnic minority group in China. The Yao minority classification has approximately 3.3 million members. They mostly live in the mountains of southern China, and approximately 60 percent live in Guangxi Province. The present study was conducted in Bama County, which is known as Yao Autonomous County in Guangxi. In primary schools across the region, Mandarin Chinese serves as the primary language of instruction for both Han and Yao students. However, the Yao dialect is also used as a supplementary language for teaching Yao students. One possible reason for this is that Yao students first encounter Mandarin Chinese only upon entering primary school, whereas Han students are exposed to it prior to starting primary school. As for English, it is a mandatory subject for both Han and Yao students at the third grade. However, Han students receive additional English instruction as part of their curriculum in grades 1 and 2, totaling approximately 4 h per week. On the other hand, Yao students only receive 2 h per week of English instruction in grades 1 and 2 due to requirements that they have to learn their own home language, Yao. After entering grade 3, the teaching hours for Han students remained unchanged at 4 h, while Yao students continued with a consistent 2-h teaching schedule.

Ethnicity and ethnic minority learners

In the present study, the following criteria were used to operationalize “ethnicity” and recruit ethnic minority learners. First, the participants’ identity was categorized as Yao in the Chinese household registry. Second, the language known by the participants before primary school was the Yao language, which is a Mienic dialect. The data were from their self-reports. We followed the participants from third grade to sixth grade. The participants were originally recruited from two primary schools, one with primarily Han students and the other with primarily Yao students. Mandarin was the main medium of instruction for students in the two schools. The original sample was 250 students. Attrition from the sample was low considering that this was a long-term study. A total of 223 students with complete longitudinal data were thus included. The students who completed all tests each year were included as participants would receive a gift coupon as a token of appreciation.

Measures

The students were assessed annually at the end of their third, fourth, fifth, and sixth grades. The instruments were individually administered to the students on each occasion. The students underwent annual assessments at the culmination of their third, fourth, fifth, and sixth grades. This was the main reason for our decision to commence the research at the end of the third grade.

Metacognitive knowledge (MCK)

In the present study, MCK was assessed with prompt-oriented individual interviews. Learners were exposed to a series of verbally and pictorially presented tasks focusing on their knowledge of cognitive processes, i.e., the ability to remember, understand, and learn. Each cognitive process entailed eight tasks. The focus of the MCK test was on people, i.e., how individuals approach various tasks in various situations; tasks, i.e., task features and processing demands; and strategies, i.e., the techniques and skills used in various tasks and contexts (Efklides & Vlachopoulos, 2012). In each task, the learners were given two or three pictures. Each picture depicted a situation in which a little boy or a girl was expected to remember, understand, or learn something. The 24 tasks were based on previous MCK studies (Annevirta & Vauras, 2001; Teng, 2022; Teng & Zhang, 2022). Figure 1 shows one sample picture task.

Fig. 1
figure 1

An example task in the MCK test

Full size image

C8-seeking explanation.

A boy was explaining the rules of a game to a girl. The girl wondered about some of the rules and did not know what those rules were about. What was the best way for the boy to explain the rules of the game? Why did the girl know/understand the rules?

The test administrator verbally described each situation in either Mandarin or the Yao language. The young learners listened to the explanation while independently checking the drawings. The learners then selected the best choice. The learners were further asked to verbally explain the reason for selecting this picture in Mandarin or Yao language (e.g., “Why did the boy you chose explain the rules in the best possible way?” “Why, in the situation you selected, did the girl better understand the rules of the game?”) (Fig. 1). In order to gather additional data and gain insights into learners’ cognitive thinking processes, we incorporated two questions: one focused on determining the most effective approach for a boy to explain the game’s rules, and another aimed at identifying the optimal method for a girl to comprehend and grasp the rules. In some situations when the learners found it difficult to answer, the test administrator guided the learners to answer by instructing them to imagine being the little boy or girl in the given situation. The test administrator could also provide some concrete prompts to help the participants reflect on how they could better remember, understand or learn something. The prompts, which were in either Mandarin or the Yao language, were as follows: (memory) “How would you try to remember as much information as possible from this activity?”, (comprehension) “How would you try to understand information for this task?”, and (learning) “How would you learn to help your classmates solve problems from this task?” The prompts were assumed to help the learners reflect on and explain the tasks.

The participants’ explanations were assessed on a three-point scale. For example, 0 points were given for not answering or giving irrelevant or simple explanations (e.g., “I think good” or “I just like it”). One point was awarded for implicit but relevant answers (e.g., “Reading a book to find details”, “I can find something new in the book”, or “I follow the teacher’s instruction to read the book”). Two points were awarded for fairly adequate and relevant answers (e.g., “It is better to explain and learn the rules by showing how the game is played”, or “You can pick up more details or rules when you try to show how the game is played”.) Three points were awarded for more explicit explanations (e.g., “Rules can be better understood by combining practice and knowledge”). The Cronbach’s alpha values for the MCK test were 0.81 (third grade), 0.85 (fourth grade), 0.83 (fifth grade) and 0.82 (sixth grade). The values ensured a sound degree of reliability.

The MCK test was conducted, accommodating participants’ language preferences by administering it in either Mandarin Chinese or the Yao language. A total of 10 English teachers, hailing from the two schools involved, took on the responsibility of administering the test, with each teacher assigned to a specific class. The 10 English teachers were newly hired and primarily responsible for teaching and administration for the first and second grade students. It is important to note that these teachers had no prior teaching experience with the participants. The research received support from the two schools, and under the guidance of the teaching department dean, the 10 English teachers generously contributed their assistance throughout the entire research endeavor. Their roles encompassed both test administration and impartial judging, with participant information being kept anonymous during the scoring process. The final score of MCK included only verbal explanations. The explanations were audio-recorded and transcribed. Three independent judges were invited to rate the participants’ verbal explanations twice. The interrater reliabilities among the three judges reached a reached an inter-rate agreement of 83–92% for the answers in the second rating. Disagreements were resolved based on majority opinions. The scores for explanations of each task were summed. The resulting maximum possible MCK score was 72 points. The scores reflected the quality of the oral explanations. The learners’ MCK was based on their explanations of their academic routine and cognitive mental processing, which, according to Annevirta et al. (2007), was consistent with Flavell’s (1979) theoretical conceptualization that metacognitive knowledge reflects learners’ comprehension, learning, and understanding with regard to persons, tasks, and strategies.

Breadth of Vocabulary knowledge (VK)

The Picture Vocabulary Size Test (PVST) by Anthony and Nation (2017) was adopted to measure children’s receptive vocabulary size and growth in learning English. The rationale of measuring learners’ receptive vocabulary size is for capturing developmental changes. Language development is a dynamic process, and receptive English vocabulary undergoes significant changes as learners progress through different stages of linguistic development. Longitudinal evaluations allow researchers to capture these developmental changes, shedding light on how vocabulary acquisition aligns with broader language development milestones. The PVST is a standardized test that can provide a quick estimate of young learners’ receptive vocabulary knowledge across childhood. The test measures whether learners can find a suitable meaning (a picture) for a given partly contextualized word form. Two 96-item test sets are included with the test. The test sets used different questions but followed the same procedures. Each image plate contained 4 pictures. The learners were required to choose one picture that best represented the meaning of the corresponding stimulus word after listening to an audio clip twice. The assessment was terminated when a child incorrectly identified six consecutive items. The measurement of vocabulary is equitable from age to age, making it appropriate for a longitudinal study. In the present study, an alternate-form reliability of 0.95 and an internal consistency reliability of 0.90 were detected, supporting the use of this test in measuring VK. The PVST yields raw scores that represent the correct number out of a total of 192 possible items. The raw score is based on the number of correct answers below the ceiling. The participants received 5 training words and image plates prior to taking the test to become familiarized with the test requirements. One common conviction upheld by second-language vocabulary researchers is that vocabulary knowledge is a multidimensional complex construct (Milton, 2009; Nation, 2001; Schmitt, 2010). However, it should be noted that the PVST only measures receptive instead of productive English vocabulary. Thus the findings in the present study could only explain the unidimensional meaning of vocabulary.

Procedure for data collection

Both the MCK and VK tests were administered one-on-one in a relaxed environment. The test administrator sat next to the child, explained the test, and kept the child motivated by providing encouraging comments such as “Good job” throughout the test. There was no fixed time limit for the tests. The MCK took approximately 40 min for each learner, while the PVST took approximately 30 min to complete. The data collection began in the second semester of the third grade. The tests were repeated annually in the second semesters of the fourth, fifth, and sixth grades. The language of instruction for all tests was in either Mandarin Chinese or Yao language. The ethnic Yao students were allowed to use either mandarin or their ethnic language to explain. All the tests were in a form of paper and pencil. One could argue that there might be a testing effect problem because the same tests were used over the course of the four years of the study. To avoid multiple testing of the same items, 10 new test items were added to each test each year. The test ceiling was not met, and in each year, adding new test items minimized the repeated testing effects in the experimental design. The new test items were intended to minimize the test–retest effects and were not included in the data analysis because the present study focused on the longitudinal development of MCK and VK.

Data analysis

Latent growth curve (LGC) analysis is a powerful technique that is based on structural equation modeling. The advantage of this technique lies mainly in its suitability for exploring participants’ individual differences in longitudinal development. The data analyses were performed using the Mplus program to evaluate longitudinal data based on the same multiple indicators at each time point (Geiser, 2013). The measurements of each variable, i.e., MCK and VK breadth, over time (from grade 3 to grade 6) were divided into several latent components to assess linear or quadratic trends, while the level/intercept component, which represents the baseline or starting point of the measurements in grade 3, are also included to provide information about the initial level of MCK and VK breadth before any changes over time. Other means, e.g., goodness-of-fit statistics, were employed to evaluate whether the data fit the models. The index values indicating the fitness of the models included the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), the goodness-of-fit index (GFI), and the comparative fit index (CFI). A cutoff value for indicating a good fit was close to 0.06 for RMSEA, 0.08 for SRMR, 0.95 for the Tucker–Lewis index (TLI) and CFI (Hu & Bentler, 1999). Data and coding for this study were shared in open science framework (https://osf.io/h7z8r/).

Results

The result section is presented based on the examined research questions.

  • Research question 1: Comparison of MCK and VK breadth between Han and Yao students

According to Table 1, the Han students seemed to have achieved higher levels of MCK than the Yao students. These findings were consistent in each year. Based on the ANOVA results, the main effect of ethnic group on MCK was significant, F = 165.565, p < 0.001, partial η2 = 0.428, indicating that the MCK scores of the Han students were significantly higher than those of the Yao students. The main effect of time was significant, F = 3356.249, p < 0.001, partial η2 = 0.938, indicating that the MCK scores increased with cumulative time. The interaction of time*ethnic group was significant, F = 353.394, p < 0.001, partial η2 = 0.615, indicating that the Han and Yao students had different development trends in the four measurements. We then conducted a repeated ANOVA to explore the variances between the Han and Yao students. The results are presented in Table 2.

Table 1 Descriptive statistical results of 4 time point measurements
Full size table
Table 2 Comparison of MCK at every point of measurement
Full size table

The repeated ANOVA in Table 2 was based on 2 (ethnic groups: Han and Yao) * 4 (time: 4 measurements) to compare MCK between the Han and Yao students. The MCK growth trend of the Han students was higher than that of the Yao students, and in each measurement, the MCK scores of the Han students were significantly higher than those of the Yao students.

The findings in Table 1 also revealed that Han students achieved higher levels of VK breadth compared to Yao students in each year (except grade 3). The ANOVA results further confirmed the significance of ethnic group as the main factor influencing VK breadth, with a substantial effect size (F = 666.507, p < 0.001, partial η2 = 0.751). These results indicate that Han students significantly outperformed Yao students in terms of VK breadth scores. Additionally, the main effect of time was also significant (F = 165.565, p < 0.001, partial η2 = 0.428), indicating that VK breadth scores increased as cumulative time progressed. The interaction between time and ethnic group was highly significant (F = 2107.641, p < 0.001, partial η2 = 0.905), indicating that Han and Yao students exhibited different developmental patterns across the four measurements. To delve into these variances between Han and Yao students, a repeated ANOVA analysis was conducted.

The results are presented in Table 3.

Table 3 Comparison of VK at every point of measurement
Full size table

Once again, a repeated ANOVA analysis was conducted, utilizing a 2 (ethnic groups: Han and Yao) * 4 (time: 4 measurements) design to compare the VK breadth between Han and Yao students. The results revealed that the growth trend in VK breadth among Han students was greater than that of Yao students. Moreover, in each measurement, the VK breadth scores of Han students were significantly higher than those of Yao students.

  • Research question 2: The longitudinal development of MCK and VK breadth in Han and Yao students

We employed LGC models to conduct a uniconstruct analysis, i.e., the associations between the initial level (i.e., intercept) of the test skill (MCK and VK breadth) and its growth over time. To achieve this goal, we conducted separate analyses based on LGC models to assess the extent to which MCK and VK breadth would be associated with the same variables of MCK and VK breadth at the uniconstruct level. The extent to which MCK and VK breadth would be associated with MCK and VK breadth refers to two growth factor components, i.e., an intercept growth factor (level) and linear growth rate (slope) estimated for each variable. We compared data fitness for the model of linear growth and the quadratic growth rate (quadratic trend). For example, we employed a linear growth model and quadratic growth model to compare whether the development trend of MCK and VK breadth should be linear growth or quadratic growth. The results for Han students are presented in Table 4.

Table 4 Comparison of data fitness for MCK and VK breadth (Han group students)
Full size table

Table 4 shows that the quadratic model is better than the linear model for MCK. This is evidence that the development of MCK for Han students is a quadratic growth model, indicating an increase in MCK over time. In terms of the development of VK breadth, the linear model is better than the quadratic model (for example, the AIC and BIC of the linear model are lower than those of the quadratic model, and the CFI and TLI are higher than those of the quadratic model). The path from the level to the linear growth of vocabulary knowledge is statistically significant. This indicates that children with initial high vocabulary knowledge can develop more vocabulary knowledge at later ages. The intercept and slope of each variable growth mode are shown in Table 5 below.

Table 5 The intercept and slope of each growth variable in Han students
Full size table

Based on Table 5, the intercept level of MCK and VK breadth at the beginning (Measurement 1) was positive and statistically significant (p < 0.001), as was the level of the slope (p < 0.001). This is evidence that there were significant individual differences in these two growth components.

The next step was to understand the development of MCK and VK breadth in Yao students. Again, we compared the data fitness for the model of linear growth and the quadratic growth rate (quadratic trend). The results are presented in Table 6.

Table 6 Comparison of data fitness for MCK and VK (Yao group students)
Full size table

Table 6 shows that the linear growth model is better than the quadratic growth model for MCK and VK breadth. The linear model better fits the MCK and VK breadth data. The intercept and slope of each variable growth mode are shown in Table 7 below.

Table 7 The intercept and slope of each variable in Yao students
Full size table

Based on Table 7, the intercept level of MCK and VK breadth at the beginning (Measurement 1) was positive and statistically significant (p < 0.001), as was the level of the slope (p < 0.001). This is evidence that there were significant individual differences in these two growth components.

  • Research question 3: The impact of MCK on VK breadth in Han and Yao students

We adopted a parallel development model based on LGC to explore the covariate relationship between MCK and VK breadth in Han and Yao students. The model mainly focuses on the relationship between the intercept and slope. The results are presented in Table 8. Although we used Generalized least squares fit by maximum likelihood (ML) at the first time, we decided to change into generalised least squares (GLS) fit by REML and bootstrapped Goodness-of-fit indices (Bollen & Stine, 1992), leading to better results in terms of model fit indices than the first time. Applying bootstrapping techniques to obtain bootstrapped goodness-of-fit indices is crucial for assessing the reliability and robustness of the model fit. Bootstrapping is a resampling method that can estimate the sampling distribution of a statistic by repeatedly sampling from the original dataset with replacement. This technique is particularly useful when the underlying distribution is unknown. Bootstrapping involves resampling the data with replacement to create multiple bootstrap samples and fitting the GLS model to each bootstrap sample and calculate the desired goodness-of-fit indices (e.g., R-squared, likelihood ratio tests, etc.). According to the cut-off value recommended in Hu and Bentler (1999), the results of using boostrapping in Table 8 reveal that the two models fit the data well.

Table 8 A summary of the model fit index of the parallel development model
Full size table

The next step was to explore the covariate relationship between MCK and VK breadth in Han students. The results are presented in Fig. 2 and Table 9.

Fig. 2
figure 2

Covariate relationship between MCK and VK breadth in Han students

Full size image
Table 9 The intercept and slope coefficient of MCK and VK breadth (Han students)
Full size table

MCK was based on the quadratic model, and VK breadth was based on the linear model to analyze the latent growth curve. As shown in Fig. 2 and Table 9, there is a significant positive correlation between the intercept of MCK and the intercept and slope of VK breadth. This is evidence that the higher the starting point for MCK is, the higher the starting point and the faster the growth rate for VK breadth in Han students.

The results regarding the covariate relationship between MCK and VK breadth in Han students are presented in Fig. 3 and Table 10.

Fig. 3
figure 3

Covariate relationship between MCK and VK breadth in Yao students

Full size image
Table 10 The intercept and slope coefficient of MCK and VK breadth (Yao students)
Full size table

The MCK and VK breadth in the Yao students were based on a linear model to explore parallel development of the latent growth curve. As shown in Fig. 3 and Table 10, there is a significant positive correlation between the intercept of MCK and the intercept of VK breadth, indicating that the higher the starting point of MCK is, the higher the starting point of VK breadth. There is a significant correlation between the linear slope of MCK and the slope of VK breadth. This shows that the faster MCK grows, the faster VK breadth grows. The two variables have a common growth trend in Yao students.

Discussion

In the present study, we examined differences between minority Yao and non-minority Han students in their development of metacognitive knowledge and breadth of vocabulary knowledge. The findings revealed significant differences between the two groups. We also investigated the individual and covariant development of metacognitive knowledge and vocabulary knowledge breadth in these young learners of grades 3 to 6. The findings support the cumulative development of students’ metacognitive knowledge and breadth of vocabulary knowledge as well as the predictive role of their metacognitive knowledge in their development of breadth of vocabulary knowledge.

Differences in Han and Yao students’ development of metacognitive knowledge and vocabulary knowledge breadth

The results suggest that although the Yao minority students and the Han students started their English learning in third grade, the Yao minority students demonstrated large, persistent deficiencies in metacognitive knowledge and vocabulary knowledge breadth. The estimates of these gaps may depend substantially on the Yao minority students’ initial level of limited metacognitive knowledge and vocabulary knowledge breadth. Even though some of the Yao minority students acquired vocabulary knowledge rapidly, by the sixth grade, they continued to lag behind the average level demonstrated by the Han students. In describing the vocabulary knowledge growth of the Yao minority students, these findings support the developmental skill-learning gap hypothesis which postulates that, as children grow older, the skill-learning gap between them and their more proficient peers widens (Wall, 2004). For these minority students, moderate deficiencies in metacognitive knowledge and vocabulary knowledge grew into large deficiencies as learning demands increased, which is consistent with previous studies (Grimm et al., 2018).

The Yao minority students who entered primary school with limited vocabulary knowledge may find it challenging to achieve further development in vocabulary knowledge breadth. This developmental process can be described as a downward spiral in regard to higher school grades because these students did not only lack vocabulary knowledge but also metacognitive knowledge, both of which were essential to their active control of vocabulary learning. Even though the Yao minority students showed progress in vocabulary knowledge learning by the end of the sixth grade, they continued to have slow rates of growth in vocabulary learning compared to the Han students. This finding supports L2 researchers’ assertion that minority students’ acquisition of English skills, including English reading and vocabulary knowledge, lags behind that of non-minority students (e.g., Kieffer, 2008). Contrarily, a separate study conducted by Choi et al. (2017) did not find any significant group differences in metalinguistic skills, nor in Chinese and English vocabulary knowledge. These findings could be attributed to the participants’ background. The participants in their study were native Cantonese-speaking children but Yao-speaking children in the present study and both English and Chinese were formally taught as language subjects starting from grade one in the school curriculum. Furthermore, the participants received eight English classes per week, which could have contributed to their similar performance in metalinguistic skills and vocabulary knowledge across both languages. The minority students’ difficulties in acquiring English vocabulary knowledge may be compounded by the limited instructional time, lack of resources, or even their lower socioeconomic status. Social-economic status has been found to have a connection with vocabulary learning, as indicated by research conducted by Hart and Risley in 1995. This relationship can be attributed to the influence of social-economic status on parent-child interactions and communication practices. However, those issues were not sufficiently explored in the present study. Such assumptions require further exploration.

Longitudinal development of metacognitive knowledge and vocabulary knowledge breadth in Han and Yao students

The findings support an upward slope with regard to the growth of MCK from primary school grade three to grade six. The quadratic MCK growth model indicates that young learners’ MCK exhibited a curvilinear trajectory, where the rate of change accelerated over time. It seems there is a non-linear relationship between time and MCK scores. The MCK scores may increase or decrease at a slower rate, then reach a turning point or inflection point where the rate of change accelerates or decelerates (e.g., from grade 4 to grade 5). The findings partially support previous studies (e.g., Annevirta et al., 2007), wherein a cumulative pattern in metacognitive knowledge from Kindergarten to Grade 2 was not detected. Consistent with previous studies (Annevirta & Vauras, 2001; Teng, 2022; Teng & Zhang, L., 2021; Teng & Zhang2022 , an earlier level of MCK affected the participants’ later development of MCK. As argued by Roeschl-Heils et al. (2003), young learners with initially low levels of MCK were not able to develop more than those who initially showed better levels of MCK. Based on the findings, we can argue that young learners in the third grade were able to master some basic skills in terms of cognitive processes, and it was possible for them to develop and reflect on their cognition, learning, and memory behaviors as their age and experience increased.

However, we can also acknowledge that the primary school third-graders were still at the beginning of their metacognitive thinking. Based on Table 1, although third-graders had begun to understand their own roles as learners in terms of taking control of cognitive activities and developed stably in their metacognitive knowledge, they began to concentrate on more mature metacognitive thinking upon entering sixth grade. Thus, the development of metacognitive knowledge may be related to grade level. As the findings showed, there were individual differences in the primary school children’s development of metacognitive knowledge (Baker, 2016; Flavell, 1979). For example, learners might begin with varied levels of metacognitive knowledge, but there are still variations in developing metacognitive knowledge after a long period of cumulative learning experience.

With regard to the linear growth model of vocabulary knowledge breadth, learners’ development followed a cumulative pattern. This is an indication that learners who acquired a certain level of breadth of vocabulary knowledge could develop such knowledge corresponding to their grade-level requirement. Thus, consistent with Sparks and Deacon (2015), learners with initially poor vocabulary knowledge might not improve more than those who initially possessed better vocabulary knowledge. With regard to breadth of vocabulary knowledge, learners at each grade level may not always develop such type of knowledge, and some learners’ breadth of vocabulary knowledge remained stable. Learners may not necessarily acquire a more pronounced breadth of vocabulary knowledge from the fifth grade to the sixth grade than from the third grade to the fourth grade. Complementing previous studies (Mancilla-Martinez & Lesaux, 2010; Teng, 2022; Zhang & Lu, 2014), there were variations in the development of vocabulary knowledge during the primary school years. One possible explanation is that such a result may be related to the role of metacognitive knowledge, which will be explained next.

The impact of metacognitive knowledge on Han and Yao students’ vocabulary knowledge breadth

The second research question examines the impact of metacognitive knowledge on vocabulary knowledge breadth. First, the findings revealed a multiconstruct relationship between young learners’ metacognitive knowledge and vocabulary knowledge breadth. Both components appeared to be highly correlated with each other. The learners who possessed higher levels of metacognitive knowledge demonstrated higher vocabulary knowledge breadth scores in each measurement. The covariant development between metacognitive knowledge and vocabulary knowledge breadth was cumulative, suggesting that metacognitive knowledge and vocabulary knowledge breadth can be developed simultaneously. The findings support those of previous studies on the role of metacognitive knowledge in predicting vocabulary knowledge (e.g., Teng, 2022). The positive findings about the relationship between metacognition and vocabulary knowledge can be attributed to the underlying mechanisms of metacognitive knowledge. Metacognitive knowledge (MK) refers to the declarative knowledge or beliefs that individuals have about the world, objects, and their own cognitive processes. It is developed through the process of abstraction, reflection, and observation of one’s own cognitive endeavors (Efklides, 2014). When it comes to vocabulary knowledge, metacognitive knowledge plays a crucial role.

In the present study, the MK test focused on learners’ knowledge of cognitive processes, i.e., the ability to remember, understand, and learn. Such metacognitive processes may allow individuals to monitor and regulate their own learning experiences, including vocabulary learning. By reflecting on their own cognitive processes, learners can gain insights into their strengths and weaknesses in vocabulary learning. Metacognitive knowledge thus enables learners to identify effective vocabulary learning strategies and adjust their approaches accordingly. For example, learners with high levels of metacognitive knowledge are more likely to engage in activities such as setting goals, planning their vocabulary learning, monitoring their progress, and evaluating effectiveness of learning strategy use as well as their overall performance. They are aware of their own vocabulary knowledge gaps and take steps to fill those gaps through targeted learning efforts. Additionally, metacognitive knowledge may help learners to self-assess their vocabulary knowledge and identify areas where they need to improve. By reflecting on their prior encounters with specific vocabulary tasks or situations, individuals can evaluate their own performance and make informed decisions on the learning strategies they should use. According to Efklides (2011), metacognitive knowledge empowers learners to actively regulate their learning processes by controlling and adapting them. The effectiveness of metacognition, specifically in terms of monitoring and control, relies on a combination of general person characteristics (such as ability, cognitive skills, metacognitive awareness, emotions, motivation, and willpower) that are applicable across various tasks and situations, as well as the specific demands and processing requirements of each task for effective self-regulation. Learners with higher levels of metacognitive knowledge are thus better equipped to manage the development of their own vocabulary knowledge, which is dependent on identifying relevant vocabulary sources, selecting appropriate learning materials, and employing effective memorization techniques.

However, while metacognitive knowledge predicted vocabulary knowledge at every time point, the acquisition of vocabulary knowledge may be influenced by various factors, including cross-linguistic factors that impact the learning of a second language (L2) vocabulary (Ecke, 2015). One significant aspect is a child’s awareness of the differences in phonological, semantic, and orthographic aspects between their first language (L1) and the L2 they are learning. When children are exposed to a new language, they may encounter phonological differences, such as distinct sounds or pronunciation patterns, between their L1 and L2. These differences can affect their ability to accurately perceive and produce L2 vocabulary. Additionally, semantic disparities arise when words in the L1 and L2 have similar meanings but are used in different contexts or have subtle variations in connotations. This can lead to challenges in understanding and correctly using L2 vocabulary. Additionally, it reflects the interplay of individual characteristics, task demands, and learning strategies involved in comprehending, learning, and memorizing vocabulary. In other words, the process of vocabulary acquisition is complex and influenced by a variety of factors. It involves a combination of metacognitive abilities and language-related resources, including phonological, morphological, and semantic aspects, and even their self-initiated and other-initiated exposure to the L1 and L2. We may need to control more variables if we aim to examine the role of metacognitive knowledge in the acquisition of vocabulary knowledge.

Limitations and implications

This study was conducted to investigate Yao and Han students’ development of metacognitive knowledge and vocabulary knowledge breadth. The findings indicate that, regardless of ethnicity, primary school students improved their metacognitive knowledge and vocabulary knowledge breadth from the third grade to the sixth grade in a cumulative fashion. However, ethnic minority students lagged behind majority Han students in developing metacognitive knowledge and vocabulary knowledge breadth. Metacognitive knowledge predicted the development of vocabulary knowledge breadth throughout the school years.

Despite the significance of this study, we must note some limitations. First, this study utilized only a single measure to assess vocabulary knowledge. The findings may not support the assessment of multi-faceted nature of vocabulary knowledge. Future longitudinal research should consider assessing both the breadth and depth of vocabulary knowledge. Such data could provide a more complete picture of students’ vocabulary learning achievements. Second, this study did not explore the participants’ socioeconomic status, which may influence young learners’ language learning development. Third, due to the nature of a longitudinal study, multiple and repeated tests were required. One might suggest that the learners may have become familiarized with the test items. Although the findings did not show a floor or ceiling effect, the learners may have acquired some items as a result of their exposure to the test once per year. Fourth, the first language for the participants was not the same. We should pay more attention to research on crosslinguistic constraints. For example future studies can explore how first language status may influence (directly/indirectly) on second language development. Finally, vocabulary knowledge may be a predictor of metacognition, particularly when the metacognitive knowledge test is heavily language dependent. This issue was not explored in the present study. In addition, more variables, including learners’ working memory, motivation, should be controlled.

Despite the limitations, this study contributes to an understanding of second language acquisition during the primary school grades for minority learners. The findings highlight the importance of developing metacognitive knowledge and vocabulary knowledge breadth for minority learners. Metacognitive knowledge is of great importance for enhancing minority and non-minority learners’ vocabulary knowledge breadth. The findings suggest that students with limited metacognitive knowledge are at risk of falling behind in their acquisition of vocabulary knowledge breadth. In particular, the findings illustrate that minority students experience greater risk in this regard. We thus argue for the instructional need for supporting young learners’s metacognitive knowledge and thus deeper understanding of L2 vocabulary. These phenomena deserve attention from bilingual or multilingual researchers and language policy makers within China and beyond.