Assessing vocabulary and grammar development in New Zealand Mandarin- and Cantonese-speaking children: a validation study

ABSTRACT

New Zealand has a growing population of Mandarin and Cantonese speakers, including children with speech, language, and communication disorders. However, the absence of language development profiles poses challenges in effectively identifying and addressing these concerns. This study aimed to capture the current state of monolingual, bilingual, and multilingual Mandarin- and Cantonese-speaking children living in New Zealand. Over 200 mothers in the Growing Up in New Zealand longitudinal cohort study identified their children as understanding Mandarin and/or Cantonese. Mothers completed newly adapted vocabulary and grammar checklists for New Zealand Mandarin and Cantonese speakers at the Age 2 data wave. Both of the adapted New Zealand Mandarin and Cantonese versions showed high reliabilities and validities. Unique demographic predictors of children’s vocabulary and grammar were mothers’ education, household deprivation level, and children’s birth order. Language status and maternal concerns were also unique predictors of children’s vocabulary development in Mandarin and Cantonese, with monolingual children whose mothers reported no concerns having higher Mandarin and Cantonese vocabularies. These results have rich implications for researchers, clinicians, and practitioners working with Mandarin- and Cantonese-speaking children in New Zealand and worldwide.

KEYWORDS:

• Bilingualism
• multilingualism; Chinese language acquisition
• assessment

Introduction

Oral language development in early childhood is a key focus for professionals and researchers as it serves as a strong predictor of later language and literacy development (Greenwood et al., 2017; Harris, Law, & Roy, 2005). The considerable impact of lexical and grammatical development in the early years on later language growth generates great interest among researchers and clinicians. For instance, children who experience language difficulties in later childhood often exhibited risk factors as toddlers (Horwitz et al., 2003; Reilly et al., 2009). Conversely, children who possess advanced oral language skills during their early years demonstrate better early literacy abilities when they enter school and enhanced reading comprehension later in primary school (Suggate, Schaughency, McAnally, & Reese, 2018). Therefore, understanding key factors influencing early language development is crucial for effective interventions and addressing long-term challenges associated with oral language issues for literacy, academic performance, and social interactions (Simonsen, Kristoffersen, Bleses, Wehberg, & Jørgensen, 2014).

In addition to the significant impact of adult input (Hart & Risley, 1995), various factors such as gender, socioeconomic status (SES), birth order, maternal education, and parental concerns have been identified as influential in shaping early language development (Horwitz et al., 2003; Reese et al., 2018; Reese & Read, 2000; Reilly et al., 2007). Furthermore, the role of bilingualism and multilingualism, the ability to speak two or more languages, has also been recognized as a significant factor in early language growth (Hoff et al., 2012; Pearson, Fernández, & Oller, 1993; Reese et al., 2018). Although a considerable body of research on Chinese children's language development has primarily been conducted in China or Chinese-dominant countries (Hao, Shu, Xing, & Li, 2008; Hua & Wei, 1999; Tardif, Fletcher, Zhang, Liang, & Zuo, 2008b), there has been an increase in studies focusing on this population in English-dominant societies in recent years (Hao, Bedore, Sheng, & Peña, 2019; Jia & Paradis, 2015; Sun, Low, & Chua, 2022; Sun, Yin, Amsah, & O’Brien, 2018). In New Zealand, there has been a noticeable increase in the population of Mandarin and Cantonese speakers, including children with speech, language, and communication disorders (Ballard & Lee, 2016). Despite this growing demographic, there is a critical gap in the identification and assessment of these children's language development within the specific context of New Zealand. The primary objective of this study was to provide a current overview of Mandarin and Cantonese acquisition among children participating in the Growing Up in New Zealand study. Specifically, we investigated the predictors of variation in language proficiency among Mandarin and Cantonese-speaking children in New Zealand with a newly adapted version of the MacArthur-Bates Communicative Development Inventory (CDI): short form (Fenson et al., 2000), with permission from the CDI board. To achieve this, we assessed 236 Mandarin and Cantonese speakers from the Growing Up in New Zealand study, a longitudinal birth cohort study that included over 6,500 diverse families representing the New Zealand population (Morton et al., 2012, 2013).

Measuring early language in bilingual children

Accurately assessing the linguistic development of young bilingual children in child language research is challenging due to the risk of over- and under-identifying speech, language, and communication disorders (Aldawood, Hand, & Ballard, 2023). Collecting, transcribing, and analysing naturalistic language samples requires extensive time and effort, which is further complicated when dealing with bilingual speakers. To overcome this challenge, parent-report inventories, such as the MacArthur-Bates Communicative Development Inventories (CDIs), have emerged as reliable and valid alternatives for assessing language proficiency in children aged 8–37 months (Fenson et al., 2000; Fenson et al., 2007; Fernald, Perfors, & Marchman, 2006). Originally developed for American English (Fenson et al., 1994, 2000), the CDI has since been adapted to over fifty languages, including different variations of English (e.g., British English: Dionne, Dale, Boivin, & Plomin, 2003; New Zealand English: Reese & Read, 2000; see https://mb-cdi.stanford.edu/adaptations.html for a current list) as well as numerous other languages (e.g., Spanish: Jackson-Maldonado, Marchman, & Fernald, 2013; Danish: Bleses et al., 2008; Canadian French: Poulin-Dubois, Graham, & Sippola, 1995). These adaptations allowed for broader applicability and enabled researchers to assess language development in diverse linguistic contexts.

The CDI:II for toddlers focuses on assessing vocabulary and grammar. Parents or caregivers complete the questionnaire by indicating the words and/or sentences their children are producing (Fenson et al., 1994). They also answer a question about children's acquisition of word combination skills, which is commonly used as an indicator of grammatical development (Bates & Goodman, 1997). The CDI:II long form provides a valid and reliable measure of language development for children aged 16–30 months but can be time-consuming, taking 30–45 min to complete. In contrast, the CDI:II short form offers a more concise version with a 100-word vocabulary checklist and one grammar question, which is also valid and reliable (Fenson et al., 2000, 2007). The short form serves as a practical alternative to the long form, particularly in situations where time and resources are limited. It has gained wide acceptance and is frequently used when administering the long form is challenging or time-consuming (Fenson et al., 2000). Therefore, the CDI short form is well-suited for comprehensive health and development studies like Growing Up in New Zealand (Reese et al., 2015, 2018).

Researchers in other countries have created several parental questionnaires tailored to assess the early language development of Mandarin- and Cantonese-speaking children (Hao et al., 2008; Hua & Wei, 1999; Tardif et al., 2008b). Tardif and colleagues adapted the CDI short forms to Mandarin and Cantonese languages to examine vocabulary and grammatical development in children aged 8–30 months from Beijing and Hong Kong, providing valuable insights into language development in Mandarin- and Cantonese-speaking populations (Tardif et al., 2008b; Tardif, Fletcher, Liang, & Kaciroti, 2009). However, Hao et al. (2008) raised concerns about the suitability of certain words on the Mandarin CDI short form, as they were direct translations from English and not colloquial to native Mandarin speakers. To address this limitation, Hao et al. (2008) developed their own Chinese vocabulary inventory and reported its reliability. Indeed, these adapted language measures were conducted in China, where most participants were monolingual and from single-child households. Most New Zealand Chinese families, however, are bilingual or multilingual with multiple children. Therefore, there is a need for adapted Mandarin and Cantonese CDIs for the New Zealand population. In this study, two novel NZ CDI:II short forms were developed exclusively for NZ Mandarin and Cantonese speakers, adapted from Tardif et al. (2008b) and Fenson et al. (2000).

Demographic factors in children’s early language development

When assessing language development in Mandarin or Cantonese speaking children in an English-speaking country, it is fundamental to consider existing research on language development in English-speaking children. Previous studies have suggested that gender can be a predictor of early lexical and grammatical development, with young girls often exhibiting more advanced linguistic skills compared to boys (Fenson et al., 2000; Reese & Read, 2000; Reilly et al., 2007). However, it should be mentioned that this association does not consistently hold true in all cases (see exceptions in Horwitz et al., 2003). More robust predictors of language development are family SES and maternal education; children from low-SES families and with mothers who have lower levels of education are more likely to have lower language skills and experience language delay (Hart & Risley, 1995; Horwitz et al., 2003; Reilly et al., 2007). Several studies have indicated that birth order can also influence language development. First-born children tend to have larger vocabularies and more advanced syntax compared to later-born children (Hoff-Ginsberg, 1998; Reese et al., 2018; Reilly et al., 2009). On the other hand, later-born children often demonstrate more advanced pronoun use and conversational skills (Hoff-Ginsberg, 1998; Oshima-Takane, Goodz, & Derevensky, 1996). These birth order effects have been linked to differences in maternal language input, with mothers using longer utterances with first-born children and employing more questioning techniques with later-born children (Hoff-Ginsberg, 1998). Finally, several studies have consistently demonstrated that parental concern (see Appendix 3 for an example), measured by a question asking mothers whether they have concerns about their children’s speech or hearing, serves as a reliable predictor of children's language development (Klee, Pearce, & Carson, 2000; Reese et al., 2018).

Research on Chinese children's language acquisition has found that girls, first-born children, those from higher socioeconomic backgrounds, and those with highly educated parents tend to show language advantages (Tardif et al., 2009; Zhang, Jin, Shen, Zhang, & Hoff, 2008). In the current study, we aimed to investigate whether similar language disparities associated with gender, socioeconomic status (area-level deprivation), language status (monolingual/bilingual/multilingual), birth order, maternal education, and maternal concern can be observed among Mandarin- and Cantonese-speaking children in New Zealand. Given that the majority number of children in each sample used English as an additional language (67% in Mandarin, 71% in Cantonese), we also examined language differences in English vocabulary and word combinations within the English-Mandarin¹ and English–Cantonese² subsamples as a function of these demographic predictors.

Bilingual and multilingual language development

Recent studies focusing on bilingual and multilingual children's language development have contributed to a better understanding of their developmental patterns and trajectories (De Houwer, Bornstein, & Putnick, 2014; Grech & McLeod, 2012). Previous research has yielded mixed findings when comparing bilingual and monolingual children in terms of vocabulary and grammatical development pace (Hoff et al., 2012; Pearson et al., 1993). It is now established that bilingual children may take longer to acquire each language compared to monolingual children (Hoff et al., 2012). However, this does not necessarily imply that bilingual or multilingual children experience language delays compared to monolingual children (Grech & McLeod, 2012). Although bilingual children may have smaller vocabularies in each language individually, their combined vocabularies (De Houwer et al., 2014) and grammatical development in any language are comparable (Hoff et al., 2012). Some researchers argue that the total number of concepts for bilingual children is similar to monolingual children when considering their languages together (Pearson et al., 1993). This conceptual vocabulary approach combines bilingual children's vocabulary knowledge across languages, taking advantage of their ability to respond in either language (Peña, Bedore, & Kester, 2016). Bilingual children are estimated to have a comparable ‘conceptual vocabulary’ when considering words across both languages without counting cross-language synonyms twice (Reese et al., 2018). However, it remains unclear how conceptual vocabulary differs from word forms. This could be because acquiring concepts and word forms varies among languages (Matthews, 1991). For example, English word form changes for tense, whereas Mandarin maintains consistent word forms regardless of grammar. Interestingly, Australian Mandarin-English bilingual children aged 3–7 showed balanced development in both languages, unaffected by language input and dominance (Yang, Kim, Tuomainen, & Xu Rattanasone, 2022).

Although there is limited research on multilingual children's speech and language development, it follows similar patterns to that of bilingual children (Grech & McLeod, 2012). It is suggested that multilingual children may take longer to reach language milestones compared to bilingual children, as their language input is distributed across three or more languages (Mieszkowska et al., 2017). To gain a comprehensive understanding of bilingual/multilingual children's development, it is crucial to compare measures of conceptual vocabulary and expressive vocabulary, along with grammatical measures, in each of their languages whenever feasible.

The present study

The emergence of Chinese immigrants in New Zealand can be historically situated in the mid-1860s. Since then, there has always been a consistent number of Chinese populations residing in New Zealand (Ballard & Lee, 2016). The 2018 census reported that a total of 247,700 Chinese individuals currently inhabit in New Zealand, constituting approximately 4.9% of the overall populace (Statistics New Zealand, 2018).

Early Chinese immigrants predominantly originated from the Canton regions of China, where Cantonese serves as their primary language. However, with the implementation of the Immigration Act 1987, there has been a 23.6% increase in Mandarin speakers over the past few decades. Many subsequent immigrants from various parts of China exhibit bilingual or multilingual abilities (Ballard & Lee, 2016). Indeed, Mandarin and Cantonese are distinct languages, lacking mutual intelligibility due to differences in their phonological systems and lexical structures (Tardif et al., 2009). One phonological contrast lies in the number of tones, with Mandarin having four and Cantonese having six. Additionally, Mandarin and Cantonese share less than half of their vocabulary, indicating significant lexical divergence (Snow, 2004). Findings from the nine-month datawave in Growing Up in New Zealand study indicated that among 6,383 mothers, 3.3% spoke Mandarin and 1.3% spoke Cantonese to their infants. By age two, 40% of the children in the full sample demonstrated comprehension of two or more languages, and 7% exhibited comprehension of three or more languages (Morton et al., 2014). Hence, the prevalence of bilingualism and multilingualism among the present generation of children is a prominent feature, highlighting the importance for professionals to update their knowledge about these children to effectively support their needs (Aldawood et al., 2023).

To the best of our knowledge, there has been no research on Chinese children’s language development from birth to two years old in a longitudinal cohort sample in New Zealand. This study focuses on Mandarin- and Cantonese-speaking children recruited through the Growing Up in New Zealand study cohort at age two. The study aimed to (1) validate the adapted New Zealand Mandarin and Cantonese versions of the CDI, and (2) explore the predictors of language acquisition in a representative group of toddlers. We hypothesized that factors such as gender, socioeconomic status (area-level deprivation), maternal education, birth order, language status, and maternal concerns would influence children's language performance. Consistent with previous research (Bleses et al., 2008; Harris et al., 2005; Horwitz et al., 2003; Jackson-Maldonado et al., 2013; Reese et al., 2015, 2018; Reese & Read, 2000; Reilly et al., 2009), we hypothesized that children from less deprived areas, with more educated parents, female, first-born, monolingual, and with no maternal concerns about speech and hearing would exhibit higher scores on the Mandarin and/or Cantonese CDI inventories. For the children who also spoke New Zealand English, we had no specific hypotheses since this study was the first to assess New Zealand English language acquisition for Chinese speakers.

Method

Participants

Growing Up in New Zealand is a comprehensive longitudinal pre-birth cohort study that involved 6,853 children and their families, representing a broad and diverse sample in terms of ethnicity and socioeconomic status (Morton et al., 2013). Recruitment of mothers and their partners (if available) occurred between 2009 and 2010, prior to the birth of their children (Morton et al., 2013). When children were approximately 24 months old (Age two data wave), approximately 196 children who spoke Mandarin and/or Cantonese and had mothers capable of reporting on their vocabulary in those languages were included in the sample, out of a total of 6,327 mother respondents. Among the overall mother respondents, 196 (3%) reported that their child could understand Mandarin, and 71 (1%) reported that their child could understand Cantonese. Further analysis was conducted on 177 (90%) of the children whose mothers could complete the vocabulary checklist in Mandarin, excluding three subsequent births of twin pairs, and on 59 (83%) of the children whose mothers could complete the checklist in Cantonese without the need for an interpreter or interviewer. Among the Mandarin-speaking sample, 20 (11%) of the children spoke Cantonese and 119 (67%) of the children also spoke English. Among the Cantonese-speaking sample, 20 (34%) of the children spoke Mandarin and 42 (71%) of the children also spoke English.

The broader Growing Up in New Zealand study collected additional demographic information relevant to language development, including birth order, maternal self-prioritised ethnicity, area-level deprivation, maternal education, and birthplace. These data were gathered during the antenatal period or through a phone call at 6 weeks, along with child gender information. Socioeconomic status was determined using the well-established 2006 New Zealand Index of Deprivation, which utilizes eight socioeconomic dimensions derived from census data developed (Salmond, Crampton, & Atkinson, 2007). Scores on this scale range from 1 (least deprived) to 10 (most deprived), and a tripartite classification of deprivation levels was used: low (≤ 3), medium (4–7), and high (8–10). Maternal education was categorized into five subgroups based on the highest level of education completed: intermediate, high school, trade certificate/diploma, bachelor's degree, or postgraduate degree. Table 1 presents the demographic characteristics of the final sample of Mandarin- and Cantonese-speaking children.

Table 1. Demographics for children and families responding to the NZ Mandarin CDI-sf and the Cantonese CDI-sf.

	Mandarin speakers N = 177	Cantonese speakers N = 59
Maternal education (highest level)
High school or lower^a (%)	10	13
Diploma/Trade certificate (%)	23	30
Bachelor's degree (%)	46	38
Postgraduate degree (%)	21	17
Maternal birthplace (% of mothers born outside NZ)	98	90
Maternal self-prioritised ethnicity (% Asian)	98	95
Area-level deprivation
Most deprived^b (%)	25	15
Moderately deprived (%)	51	53
Least deprived (%)	24	30
Child gender (% male)	49	52
Birth order (% first born)	61	55
Child language status
Monolingual (%)	31	23
Bilingual^c (%)	52	42
Multilingual^d (%)	18	35

Notes: ^aThe value of this variable for the Cantonese sample is fewer than 10.

^b The value of this variable for the Cantonese sample is fewer than 10.

^c,d Among the bilingual and multilingual children, 67% of the Mandarin speakers and 71% of the Cantonese speakers used English as another language.

Measures

Expressive language

Child vocabulary in Mandarin and Cantonese and word combination skills in any language were measured using the two new adapted versions of the MacArthur-Bates Communicative Development Inventory (CDI) short form in Mandarin and Cantonese for New Zealand children. For children who also spoke English, their English vocabulary was also assessed with the NZ CDI:II short form (Reese et al., 2018). To construct the New Zealand adapted versions of the CDI for Mandarin and Cantonese, the third and fourth authors, who are Mandarin and Cantonese native speakers respectively and had experience working with children, consulted with Chinese families more generally and made recommendations to Tardif’s (2008b) existing word list. These were discussed with the second author before items that did not fit the NZ context were removed to produce 100-word versions that were similar in scope to the long-form Chinese versions (Tardif et al., 2008b) regarding lexical categories, such as action words, pronouns etc. Each word was assigned a score of one point, resulting in a scoring range from 1 to 100 for the questionnaire. Additionally, the questionnaire included a question regarding the children's ability to combine words in any language for a measure of grammatical development. The response options for this question were ‘not yet’ ‘sometimes’ and ‘often’; the scoring range for this question was from 1 to 3.

Procedure

Antenatal measures

During the antenatal phase, mothers were interviewed using a 90-minute face-to-face computer-assisted personal interview (CAPI) conducted by trained interviewers at the family's home. The interview covered a range of topics, including questions about the mothers themselves, their families, and other important information such as the household SES relevant to their children.

Child expressive language at two years

At around 24 months of age, the child expressive language assessment took place during the computer-assisted personal interview (CAPI) with Mandarin- and Cantonese-speaking mothers as part of the Growing Up in New Zealand study. Initially, mothers were asked to indicate all the languages that their children could understand, including New Zealand English. Subsequently, for each of these languages, mothers were presented with show cards containing numbered word lists specific to the respective language (see the Appendices for the word lists in Mandarin and Cantonese; for the word list in New Zealand English, please refer to Reese et al., 2018). The standard instructions for the CDI: II short form (Fenson et al., 2000) were slightly adapted for the computer-assisted procedure: Children understand many more words than they say. With this question, we are particularly interested in the words your child SAYS. Please look at this list and tell me the numbers for the words you have heard [Child’s name] use. If [Child’s name] uses a different pronunciation of a word (e.g., ‘sketti’ instead of ‘spa- ghetti’), still tell me the number for that word. Remember that this is a list of all the words that are used by many different children – don’t worry if your child only uses a few of these words (Reese et al., 2018). Interviewers entered the participant’s response for each numbered word on each show card.

Following the completion of the vocabulary checklist in all the languages the mother believed her child understood, the word combination question was asked in English (Has your child begun to combine words yet, such as ‘another cookie’ or ‘doggie bite’?; Fenson et al., 2000), provided with three options: ‘not yet’ ‘sometimes’ and ‘often’. For this particular question, mothers were asked to provide their response considering all languages spoken by the child.

After completing the word combination question, mothers were asked if they had any concerns about their child's speech or hearing. If they expressed a concern, they were further prompted to specify the type of concern using the categories provided in Appendix 3. Additionally, mothers were asked to rate the level of their concern, ranging from not concerned at all to very concerned. Furthermore, they were asked whether they had sought professional advice or treatment regarding their child's speech and/or understanding.

Results

Psychometric properties

To evaluate the reliability of each inventory, Cronbach's alpha was employed as a measure of inter-item correlations within each scale. The Mandarin and Cantonese vocabulary scales demonstrated high reliability, with Cronbach's alpha coefficients of .99 for both. These high alpha values are consistent with those reported for the English CDI: II short forms (Fenson et al., 2000).

Vocabulary development

Table 2 displays the average vocabulary scores for Mandarin and Cantonese speakers as a function of by gender, birth order, language status (monolingual, bilingual, and multilingual), maternal education, area-level deprivation (least, moderately, and most deprived), and maternal concerns regarding the child's speech or hearing. Language status was dummy coded as monolingual, bilingual, and multilingual; area-level deprivation was dummy coded into least, moderately, and most deprived. Prior to regression analysis, the intercorrelations among the predictors were assessed, revealing low correlations among all eight predictors (r_s < .6); therefore, all were retained and included in the final regression model. Due to minimal variation in maternal self-prioritised ethnicity and maternal birthplace, these two variables were excluded from the regression analysis. Specifically, 98% of Mandarin and 95% of Cantonese participants identified themselves as Asian, while 98% of Mandarin and 90% of Cantonese participants were born outside of New Zealand. Multiple linear regression analyses were conducted to explore the best predictors for children’s vocabulary knowledge. Two separate regression models were built for the Mandarin and Cantonese samples and two separate regression models for the English-Mandarin and English–Cantonese subsamples. Percentile rankings for monolingual, bilingual, and multilingual Mandarin- and Cantonese-speaking children on both inventories can be found in Appendices 4–6.

Table 2. Mean vocabulary (and SD) and word combination scores on the NZ Mandarin and Cantonese CDI-sf as a function of demographics and maternal concerns.

	Mandarin vocabulary (total words)	Word combinations (% not yet combining)	Cantonese vocabulary (total words)	Word combinations (% not yet combining)
Child gender
Girls	29.69 (29.15)	10	32.79 (32.47)	3
Boys	28.16 (27.85)	20	24.20 (24.21)	10
Birth order
First-borns	31.88 (30.53)	11	26.40 (27.62)	6
Later-borns	24.44 (24.47)	20	31.00 (30.25)	8
Language status
Monolinguals	52.65 (31.77)	13	59.23 (30.40)	0
Bilinguals	20.80 (19.61)	15	22.92 (21.53)	8
Multilinguals	11.77 (16.71)	16	15.90 (21.22)	10
Maternal education^a
Less educated	32.17 (27.93)	12	38.73 (30.98)	9
More educated	27.32 (28.68)	16	20.30 (24.16)	4
Area-level deprivation
Most deprived	30.25 (28.72)	16	32.78 (33.35)	11
Moderately deprived	26.81 (26.84)	13	28.44 (28.38)	10
Least deprived	32.05 (31.60)	16	26.22 (28.17)	0
Maternal concern
Not concerned	31.38 (29.62)	11	31.04 (29.00)	6
Concerned	18.68 (20.14)	30	11.75 (20.67)	13

Notes: ^aLess educated = highest level of education intermediate or high school qualification; more educated = highest level of education trade certificate/diploma or bachelor’s or postgraduate degree.

Mandarin speakers

For Mandarin speakers, the average total vocabulary was 28.94 words. The most common words were 鞋 (shoe, 65%), 水 (water, 64%), 狗 (dog, 64%), 手 (hand, 61%), 抱 (hug, 53%) (see Appendix 1). When all the predictors were entered simultaneously into a regression analysis, the model accounted for 36% of the variance in Mandarin vocabulary, R²= .36, F(8, 168) = 11.62, p < .001, which could be explained by two unique predictors in the final model: monolingual (β = 0.49, p < .001) and maternal concerns (β = −0.16, p = .01). Mandarin monolingual children and those whose mothers had no concerns about children’s speech and hearing exhibited larger vocabularies than Mandarin bilingual children or those whose mothers had concerns.

English-Mandarin speakers

For English-Mandarin speakers, the average total English vocabulary was 19.56 words. When all the predictors were entered simultaneously into a regression analysis, the model accounted for 13% of the variance in English vocabulary, R²= .13, F (7, 111) = 2.41, p = .02. Maternal education was the only significant predictor (β = 0.28, p = .00); children whose mothers reported lower education had smaller English vocabularies than children whose mothers reported higher education.

Cantonese speakers

For Cantonese speakers, the average total vocabulary was 28.13 words. The most common words were 狗 (dog, 58.3%), 車 (car, 55%), 打 (hit, 53.3%), 頭髮 (hair, 50%), 襪 (sock, 48.3%), 帽 (hat, 48.3%) (see Appendix 2). When all the predictors were entered simultaneously into a regression analysis, the model accounted for 47% of the variance in the Cantonese vocabulary checklist, R²= .47, F(8, 50) = 5.49, p < .001, which could be explained by three unique predictors in the best fitting model: maternal education (β = −0.29, p = .01), monolingual status (β = 0.44, p < .001), and maternal concerns (β = −0.24, p = .03). Cantonese monolingual children, those whose mothers reported less education, and those whose mothers reported no concerns exhibited larger vocabularies than Cantonese bilingual children, those whose mothers reported higher education, and those whose mothers had concerns about the child’s speech or language.

English–Cantonese speakers

For English–Cantonese speakers, the average total English vocabulary was 20.26 words. When all predictors were entered simultaneously into a regression analysis, the model was not significant and there were no significant differences in English vocabulary by gender, birth order, language status (bilingual and multilingual), maternal education, area-level deprivation (least, moderately, and most deprived), or maternal concerns, R²= .20, F (7, 34) = 1.25, p = .31.

Overall, the patterns for vocabulary size for the Mandarin and Cantonese full samples as a function of demographics were similar for language status and maternal concerns: Monolingual Mandarin and Cantonese speakers and those whose mothers reported no concerns had larger vocabularies compared to bilingual children or those whose mother reported concerns in that language. However, for Cantonese speakers only, vocabulary differed as a function of maternal education in the opposite direction to that predicted, with children whose mothers were less educated having a larger Cantonese vocabulary than children whose mothers were more educated. This pattern was the opposite for the English vocabularies in English-Mandarin subsamples as a function of maternal education, with children whose mothers reported lower education having smaller English vocabularies than children whose mothers reported higher education.

Grammatical development

Table 2 also displays the average word combination scores for Mandarin and Cantonese speakers as a function of gender, birth order, language status (monolingual, bilingual, and multilingual), maternal education, area-level deprivation (least, moderately, and most deprived), and maternal concerns regarding the child's speech or hearing. Again, language status was dummy coded as monolingual, bilingual, and multilingual, while area-level deprivation was dummy coded into least, moderately, and most deprived. Before conducting the regression analysis, the intercorrelations among the predictors were examined to avoid multicollinearity and found to be low (r_s < .6) for all eight predictors. Consequently, all predictors were retained and included in the final regression model. Maternal self-prioritised ethnicity and maternal birthplace were excluded from the regression analysis as most mothers of Mandarin and Cantonese speakers self-prioritised as Asian (98% of Mandarin and 95% of Cantonese) and were born outside New Zealand (98% of Mandarin and 90% of Cantonese). Since the dependent variable is discrete, multivariable ordinal regression models were used to estimate the cumulative odds ratio (OR) for word combinations with all predictor variables. Two separate regression models were built for the Mandarin and Cantonese samples and two separate regression models were built for the English-Mandarin and English–Cantonese subsamples.

Mandarin speakers

A chi-square test of model fit demonstrated that the model was a good fit to the data, χ² (8) = 25.64, p <.001. Significant variations existed in Mandarin speakers’ word combinations as a function of birth order, area-level deprivation (moderately and most deprived), and maternal concerns. The odds of children combining words in any language were 2.40 times higher (95% CI [1.27, 4.51]) for later-born children compared to first-born children, Wald χ² (1) = 7.29, p < .01. Additionally, the odds of children combining words in any language were 3.62 times lower (95% CI [1.71, 7.67]) for mothers who reported concerns compared to those who did not, Wald χ² (1) = 11.31, p < .001. Furthermore, the odds of children combining words in any language were 2.36 times higher (95% CI [1.13, 4.91]) for moderately deprived households compared to the most deprived households, Wald χ² (1) = 5.22, p = .02.

English-Mandarin speakers

A chi-square test of model fit demonstrated that the model was a good fit for the English-Mandarin subsample, χ² (7) = 26.23, p < .001. Significant variations existed in English-Mandarin speakers’ word combinations as a function of birth order, maternal education, area-level deprivation (moderately and most deprived), and maternal concerns. The odds of children combining words in any language were 2.37 times higher (95% CI [1.06, 5.28]) for later-born children compared to first-born children, Wald χ² (1) = 4.46, p = .04. The odds of children combining words in any language were 3.13 times higher (95% CI [1.26, 7.75]) for children whose mothers reported higher education compared to children whose mothers reported lower education, Wald χ² (1) = 6.07, p = .01. The odds of children combining words in any language were 4.31 times higher (95% CI [1.66, 11.19]) for children in moderately deprived household compared to children in most deprived households, Wald χ² (1) = 9.00, p <.001. The odds of children combining words in any language were 4.27 times higher (95% CI [1.68, 10.88]) for children whose mothers reported no concerns compared to children whose mothers reported concerns, Wald χ² (1) = 9.26, p <.001.

Cantonese speakers

For Cantonese speakers, there were no significant differences in word combinations by gender, birth order, language status (monolingual, bilingual, and multilingual), maternal education, area-level deprivation (least, moderately, and most deprived), or maternal concerns, Wald χ² (8) = 7.56, p = .48.

English–Cantonese speakers

For English–Cantonese speakers, there were no significant differences in word combinations by gender, birth order, language status (bilingual and multilingual), maternal education, area-level deprivation (least, moderately, and most deprived), or maternal concerns, Wald χ² (7) = 6.02, p = .54. Overall, in both the full Mandarin samples and the English-Mandarin subsample, word combinations were predicted by similar demographic factors such as birth order, maternal concerns, and area-level deprivation (moderately and most deprived). However, maternal education was a significant predictor only in the English-Mandarin subsample, and not in the full Mandarin sample.

Discussion

This study had two aims: (1) to assess the reliability of the New Zealand-adapted short forms of the Mandarin and Cantonese versions of the Communicative Development Inventories (CDI), and (2) to investigate the language development of monolingual, bilingual, and multilingual Mandarin- and Cantonese-speaking children in the context of the Growing Up in New Zealand study. We found that both of the NZ adapted Mandarin and Cantonese CDI short forms demonstrated high reliability. Furthermore, for vocabulary development, language status and maternal concerns emerged as strong predictors of vocabulary in Mandarin and Cantonese-speaking children, with monolingual children whose mothers reported no concerns exhibiting larger vocabularies compared to bilingual speakers whose mothers showed concerns. For Cantonese speakers only, children of mothers having lower education displayed larger Cantonese vocabularies.

The picture was different for Mandarin-speaking children who also spoke English. We found that English-Mandarin speaking children whose mothers reported lower education had smaller English vocabularies compared to children whose mothers reported higher education. Additionally, maternal education was a predictor for English-Mandarin speakers’ word combinations; English-Mandarin speaking children whose mothers reported lower education were less likely to combine words than children whose mothers reported higher education.

For grammatical skills across all children’s languages, similar predictors identified were birth order, SES, and maternal concerns of Mandarin-speaking and English-Mandarin speaking children's grammar, where children from households with high deprivation and whose mothers expressed concerns about their speech or hearing exhibited lower word combination scores. Contrary to expectations, later-born Mandarin speakers exhibited more advanced grammatical skills across their languages than first-borns.

Vocabulary development

The findings revealed that children in both language groups produced less than 30% of the total vocabulary listed in the inventories. Specifically, the most frequently used words among Mandarin-speaking children were ‘shoe,’ ‘water,’ ‘dog,’ ‘hand,’ and ‘hug,’ whereas Tardif et al.'s (2008a) study identified ‘mommy,’ ‘daddy,’ ‘grandma-paternal,’ ‘grandpa-paternal,’ and ‘hello/wei’ as the top words in their sample. For Cantonese speakers, the prominent words in our study were ‘dog,’ ‘car,’ ‘hit,’ ‘hair,’ ‘sock,’ and ‘hat,’ whereas Tardif et al.'s (2008a) study found ‘daddy,’ ‘aah,’ ‘mommy,’ ‘YumYum,’ and ‘older sister’ as the most common words. These variations in word types and categories can be attributed to multiple factors. Firstly, the children in our study were assessed at the age of two, which is older than the children in Tardif et al.'s (2008a) study, who ranged from 8 to 16 months old. Additionally, the differences in word selection may arise from disparities in the inventories themselves. It is interesting that the word ‘dog’ emerged as a top word for both Mandarin and Cantonese-speaking children in our study but not in Tardif et al.'s (2008a) study. This dissimilarity may be due to age differences or possibly influenced by the prevalence of dog ownership within New Zealand Chinese families. Moreover, the greater exposure to dogs in public spaces in New Zealand compared to mainland China or Hong Kong could contribute to this discrepancy. Furthermore, the frequent depiction of dogs in children's books in New Zealand, which we call the ‘Hairy Maclary effect’ after the popular book series by Lynley Dodd, could influence children's word preferences.

Our hypothesis was that children from less deprived areas, with more educated parents, female, first-born, monolingualism, and no maternal concerns about speech and hearing would score higher on the Mandarin and/or Cantonese CDI inventories. Partially aligned with this hypothesis, our result was consistent with previous research suggesting that bilingual children had lower vocabularies in one of their languages than did monolingual speakers (see also Hoff et al., 2012; Reese et al., 2015). When comparing the language performance of children in our Mandarin (N = 54) and Cantonese (N = 13) samples to other Chinese samples (i.e., see Hao et al., 2008; Tardif et al., 2009), we found that monolingual children in our study used a similar higher number of words from the two lists (Mandarin: Mean = 52.65, SD = 31.77; Cantonese: Mean = 59.23, SD = 30.40). However, this does not imply that bilingual children experience delayed language learning. In fact, bilingual children exhibit comparable language development across both languages compared to monolingual individuals (see Hoff et al., 2012).

It is important to note that the majority of children in both samples had English as an additional language (67% of Mandarin speakers and 71% of Cantonese speakers). For English-Mandarin speakers (N = 119), the total vocabularies in English (Mean = 19.56, SD = 18.42) and Mandarin (Mean = 18.21, SD = 17.94) were similar. This finding aligns with a recent study by Yang et al. (2022), which suggested no difference in language dominance and the amount of language input on bilingual children's narrative skills in each language between age 3 and 7. It is possible that the bilingual English speakers in our samples performed similarly across both of their languages because they were younger, and therefore had more exposure to all of the languages compared to Yang et al. (2022). However, we did not assess language dominance as perceived by mothers, which is a limitation of our design. Future investigations should explore the relative use and proficiency of each language within home and school environments to obtain a more comprehensive understanding of bilingual children's language acquisition.

Contrary to our hypothesis, Cantonese-speaking children whose mothers had lower education levels demonstrated larger vocabularies in Cantonese. This finding may be attributed to the tendency of mothers with lower education to overestimate their children's language skills using CDIs (Jackson-Maldonado et al., 2013). Alternatively, less educated mothers might predominantly use Cantonese as the primary language with their children due to their limited exposure to alternative languages, thus positively impacting their Cantonese vocabulary development. An additional possible explanation for this phenomenon is the increased influx of Mandarin speakers to New Zealand. Mothers with higher education likely have more language skills across Mandarin/Cantonese and English (and/or other languages) than mothers with less education, leading to higher language fluency and competence in their other languages (Ballard & Lee, 2016). This possibility is supported by the results from our English-Mandarin sample where children whose mothers reported lower education had smaller English vocabularies and word combination scores compared to children whose mothers reported higher education. Additionally, mothers with higher education may be more likely to have employment where they primarily utilize English and/or both Mandarin/Cantonese and English (Jia & Paradis, 2015). This circumstance could result in more bilingual/multilingual interactions with their children, potentially leading to a more balanced split in language skills across the child's languages. These factors could influence bilingual mothers with higher levels of education to opt to speak Mandarin/English rather than Cantonese to their children, perceiving it as a more practical and advantageous language (Ballard & Lee, 2016). Using a non-parametric test, a statistically significant disparity in the median highest maternal education level was detected when comparing the Mandarin and Cantonese samples (excluding 20 participants who spoke both languages). Specifically, mothers of Cantonese-speaking children exhibited lower educational attainment on average in comparison to mothers of Mandarin-speaking children. To better understand the complex relationship between maternal education and child language outcomes in bilingual/multilingual samples, future studies should consider asking mothers what language their education is based on at the point of data collection (see Sorenson Duncan & Paradis, 2020).

Grammatical development

For word combination scores, only 15% of Mandarin speakers and 7% of Cantonese speakers had not yet begun combining words in any language at age two. These results align with word combination scores for New Zealand English speakers (10% not yet combining; Reese et al., 2018) in the full Growing Up in New Zealand cohort. In contrast, lower rates of word combinations were observed among NZ Samoan (17% not yet combining; Reese et al., 2015) and Tongan speakers (16% not yet combining; Reese et al., 2015) in the same cohort.

In the Mandarin sample, including English-Mandarin speakers, we observed that children from the most deprived households showed lower rates of word combinations compared to children from moderately deprived households. This finding supported our hypothesis and is consistent with the word combination patterns seen in English-speaking children from the entire group (Reese et al., 2018). It also aligns with previous research indicating that children in less deprived households tend to have language advantages (Horwitz et al., 2003; Jackson-Maldonado et al., 2013; Reese et al., 2015). However, children from the most deprived households did not also have smaller Mandarin vocabularies as predicted. It is possible that these children may receive similar input in Mandarin as children in families from moderately deprived areas. However, their exposure to languages outside of Mandarin, particularly in early childhood education and other social settings, may be more limited.

Contrary to hypotheses, we found that later-born children exhibited more advanced grammar skills compared to first-born children in the Mandarin sample, including the English-Mandarin sample. This finding contradicts previous research indicating the opposite pattern (Hoff-Ginsberg, 1998; Horwitz et al., 2003; Reilly et al., 2009). One possible explanation is that later-born children benefited from increased exposure to grammatical cues through interactions with parents and with older siblings (Oshima-Takane et al., 1996). Given the number of English-Mandarin speakers, it is plausible that later-born children may exhibit a higher frequency of word combinations in English due to their exposure to older siblings who speak English, which could contribute to their advanced grammar skills compared to first-born children.

Maternal concerns

We predicted that children whose mothers had concerns with their children’s speech or hearing would be more likely to have lower language levels. Aligned with our hypothesis, maternal concerns were a unique predictor of Mandarin- and Cantonese-speaking children’s language development for vocabulary skills; those whose mothers displayed concerns about children’s speech and hearing had less vocabulary skills than those who showed no concerns. Furthermore, we found Mandarin-speakers (including English-Mandarin speakers) whose mothers expressed no concerns were more likely to combine words than children whose mothers expressed concerns. Our findings align with existing research that highlights parental concern as a significant predictor for children's language delay in New Zealand and in other countries (Harris et al., 2005; Horwitz et al., 2003; Klee et al., 2000; Reese et al., 2018), despite some previous studies failing to establish its predictive value (e.g., Dale, Price, Bishop, & Plomin, 2003). It is important to consider the potential influence of shared reporting bias, as mothers who had concerns could have underestimated their children's linguistic abilities. Additionally, it is possible that mothers with concerns intentionally used less sophisticated vocabulary and grammar when interacting with their children, thereby limiting their expressive language skills. However, this effect was not observed in our Cantonese sample, possibly due to limited statistical power. Further investigation is required to explore the quantity and quality of maternal input in cases where concerns are present, as well as their associations with child language development.

Strength, limitations, and future directions

The present study encompasses several strengths and limitations. Notably, the inventories employed demonstrated high reliability in both languages and exhibited predictive relationships with sociodemographic factors as anticipated. However, certain limitations need to be addressed. First, since we used adapted Mandarin and Cantonese CDIs to evaluate bilingual language acquisition, direct comparisons between the results obtained from the new adapted inventories and other versions for assessing children's overall vocabulary across both languages are not feasible. The two wordlists are not aligned since they were selected to include common words used in Mandarin and Cantonese speaking families (see Peña, 2007). As a result, we were unable to calculate composite scores based on the number of concepts across all languages of the children (see Pearson et al., 1993). Moreover, the study did not assess other language varieties, such as Shanghainese or Fukien, which may limit the ability to accurately determine vocabulary scores for bilingual and multilingual children who can speak Mandarin or Cantonese in addition to these other languages.

Second, we recognize that our study could have benefited from a more comprehensive collection of background information about the families, especially those where fathers or grandparents were the main caregivers. Caution should be exercised when generalizing our findings given our focus only on mothers as respondents. Since many young Chinese children are raised primarily by their grandparents, parental report measures may not reflect language acquisition accurately (Reese et al., 2015; Tardif et al., 2009). According to Morton et al. (2014), 27% of Asian children in the Growing Up cohort lived with extended families at age two. This suggests that the family dynamics in immigrant Chinese families can be vastly different from other families in the cohort, with grandparents often interacting with their young grandchildren in their native dialect rather than English. In future research, reports from all caregivers should be included for better estimates of Mandarin and Cantonese proficiency in children, and conducting ethnographic interviews will provide valuable insights into the family dynamics. Language samples of children interacting with grandparents and siblings would also round out the picture to provide a more holistic understanding of bilingual language acquisition in young Chinese-speaking children.

The findings of our study hold significant implications for clinicians in assessing the language development of Chinese-speaking children in New Zealand. Clinicians often have robust resources for assessment/intervention in English, but face barriers when working with children who speak a language other than English and/or in addition to English, particularly non-European languages. Given that more Chinese-speaking children may be turning up in practice, it is of great benefit to the profession to have assessments in Mandarin/Cantonese that have been validated for the New Zealand context. We would like to temper that by reminding our readers of the importance of collecting detailed information about language exposure/environment and taking caregiver concerns into consideration. Specifically, clinicians should take into account the language environment of the child, whether monolingual or bilingual, and to assess all of the child’s languages whenever possible with multiple respondents for a more complete picture. Ideally, bilingual and multilingual children should also be assessed with separate questions about word combinations in each of their languages. These word combination questions are vital for identifying language delays, especially in children under 24 months old (Klee et al., 2000). Finally, given our findings that maternal concerns regarding children's speech or hearing were robust predictors of both total vocabulary and word combinations among Mandarin speakers, and of vocabulary among Cantonese speakers, caregivers’ concerns need to be taken seriously. It is recommended that general practitioners, teachers, and clinicians in New Zealand inquire about caregivers’ concerns regarding speech or hearing as a preliminary screening tool prior to conducting further assessments or making referrals for Mandarin- and Cantonese-speaking children.

Conclusions

In summary, this study validated versions of Mandarin and Cantonese CDIs specifically tailored for the New Zealand context, and taking into account their proficiency in New Zealand English. The study also examined significant predictors of language acquisition among 24-month-old Mandarin- and Cantonese-speaking children residing in New Zealand. Notably, the adapted versions of the Mandarin and Cantonese CDI short forms demonstrated high levels of reliability and validity. Furthermore, the findings underscored the role of demographic factors on the language acquisition of Chinese children. It is expected that these findings provide a practical resource with insights on the development of bilingual and multilingual children, as well as for clinicians and practitioners engaged in supporting Mandarin- and Cantonese-speaking children both in New Zealand and on a global scale.

Acknowledgements

We express our gratitude to the whānau participating in the Growing Up in New Zealand study. We would also like to thank Philip Dale and the MacArthur-Bates Advisory Board for granting permission to adapt the Communication Development Inventory. The study was funded by various organizations, including the New Zealand Ministry of Social Development, Ministry of Health, University of Auckland, Social Policy and Evaluation Research Unit (Superu, formerly Families Commission), Ministry of Business, Innovation and Employment, Ministry of Education, Ministry of Justice, Sport New Zealand (previously Sport and Recreation New Zealand), New Zealand Police, Te Puni Kokiri, Housing New Zealand, Ministry of Pacific Island Affairs, Ministry of Women's Affairs, Department of Corrections, and the Mental Health Commission.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Ministry of Business, Innovation and Employment; Ministry of Education- New Zealand; Mental Health Commission - New Zealand; Ministry of Social Development - New Zealand; Ministry of Health - New Zealand; Department of Corrections - New Zealand; Ministry of Justice - New Zealand; Sport New Zealand; New Zealand Police; Te Puni Kokiri; Housing New Zealand; Ministry of Pacific Island Affairs - New Zealand; Ministry of Women's Affairs - New Zealand; Social Policy Evaluation and Research Unit; University of Auckland.

Notes

1 For this study, English-Mandarin speakers refers to bilingual or multilingual Mandarin speakers who used English as an additional language.

2 For this study, English-Cantonese speakers refers to bilingual or multilingual Cantonese speakers who used English as an additional language.

Appendices

Appendix 1. Short form of the NZ Mandarin CDI:II (NZM CDI-sf).

Table

1 喵 (猫叫) Meow	34 豆芽 Sprouts	67 云 Cloud
2 哎哟 Oh	35 水果 Fruit	68 车 Car
3 宝宝 Baby	36 梨 Pear	69 卡车 Truck
4 阿姨 Aunt	37 辣椒 Chili pepper	70 动物园 Zoo
5 弟弟 Younger brother	38 头发 Hair	71 幼儿园/托儿所 Kindergarten/Nursery
6 老师 Teacher	39 舌头 Tongue	72 中国 China
7 要 Want	40 手 Hand	73 低下 Lower down/Below
8 你好 Hello	41 膝盖 Knee	74 旁边 Next to
9 请 Please	42 狗 Dog	75 二 Two
10 (真) 棒! (So) Great!	43 动物 Animal	76 全部 All
11 坐 Sit	44 鸟 Bird	77 自己 Self
12 有 Have	45 蚊子 Mosquito	78 那个 That
13 玩 Play	46 恐龙 Dinosaur	79 我们的 Ours
14 给 Give	47 好 Good	80 人家 Others/They
15 送 Send	48 香 Fragrant	81 条 Classifier for long or thin objects
16 掉 Drop	49 烫 Hot (to the touch)	82 块 Classifier for piece or chunk
17 拍 Pat/Slap	50 可怜 Pitiful	83 位 Classifier for people
18 等 Wait	51 生气 Angry	84 些 Some
19 揉 Rub	52 够了 Enough	85 在哪儿/哪儿呢? Where is it?
20 翻 Flip/Turn over	53 小心 Be careful	86 哪个 Which
21 摘(楸) Pick	54 碗 Bowl	87 多少 How many
22 记(得) Remember	55 电扇 Electric fan	88 嘛 Modal particle used to express obviousness
23 用 Use	56 东西 Things/Stuff	89 噢 Okay
24 游泳 Swim	57 球 Ball	90 今天 Today
25 套 Cover	58 棋子 Chess piece	91 已经 Already
26 躲 Hide	59 背心 Vest	92 以前 Before
27 问 Ask	60 上衣 Top (clothing)	93 能 Can
28 假装 Pretend	61 鞋 Shoes	94 肯 Willing to
29 希望 Hope	62 床 Bed	95 不许 Not allowed
30 水 Water	63 抽屉 Drawer	96 跟 With/Follow
31 粥 Porridge	64 花 Flower	97 因为 Because
32 饺子 Dumplings	65 树枝 Branch	98 抱 Hug
33 肉 Meat	66 棍子 Stick	99 看 Look/See/Watch
		100 凳子/椅子 Stool/Chair

Appendix 2. Short form of the NZ Cantonese CDI:II (NZC CDI-sf).

Table

1 喵 (貓叫聲) Meow	34 知道 Know	67 天(空) Sky
2 呷 (鴨叫聲) Quack	35 試 Try	68 海 Sea
3 媽媽/媽咪Mom/Mommy	36 送 Send	69 襪 Socks
4 爸爸/爹地 Dad/Daddy	37 揭(開) Open	70 帽 Hat
5 嬸嬸 Aunt	38 多謝/謝謝 Thank you	71 (衫) 鈕/扣 Button (on a shirt)
6 妹妹 Younger sister	39 唔該 Please	72 拉鍊 Zipper
7 哥哥 Older brother	40 刷牙 Brush teeth	73 頭髮 Hair
8 司機 Driver	41 洗手 Wash hands	74 脷 Tongue
9 水 Water	42 食飯 Eat meal	75 背脊 Back
10 粥 Porridge	43 打電話 Make phone calls	76 床 Bed
11 腸粉 Rice noodle roll	44 波 Ball	77 櫃捅 Cabinet
12 蝦 Shrimp	45 積木/Lego Building blocks	78 洗衣機 Washing machine
13 豆腐 Tofu	46 故仔/故事 Story	79 屋企 Home
14 蘋果 Apple	47 冇 Don’t have	80 多 More
15 雪糕 Ice cream	48 得意/可愛 Cute	81 一樣 Same
16 吐 (luer) Spit	49 肉酸/核突 Sour	82 嗰邊/嗰度 There
17 打 Hit	50 凍 Cold	83 前面 Front
18 俾 Give	51 頸渴/口渴 Thirsty	84 我嘅 My
19 攞 Take	52 滑 Slip	85 佢 He/She
20 抱 Hold	53 快 Fast	86 我地 We
21 行(路) Walk	54 黑(色) Black	87 (一) 個 Classifier for object
22 學 Learn	55 狗 Dog	88 (一) 隻 Classifier for animals
23 熄(燈) Turn off (light)	56 蛇 Snake	89 (一) 本 Classifier for books
24 用 Use	57 熊仔/啤啤 Teddy bear	90 (一) 張 Classifier for flat objects
25 喊 Shout	58 蝴蝶 Butterfly	91 乜嘢 What
26 問 Ask	59 車 Car	92 得唔得? Is it okay?
27 唱 Sing	60 飛機 Airplane	93 啦 Final particle to soften the tone
28 睇 Watch	61 碗 Bowl	94 喎 Final particle to express agreement
29 聽 Listen	62 番揀 Choose	95 想 Want
30 梳(頭) Comb (hair)	63 鎖匙 Key	96 鍾意 Like
31 摸 Touch	64 垃圾 Garbage	97 依家 Now
32 拋 Throw	65 樹 Tree	98 先 First
33 著(衫) Wear (clothes)	66 遮(雨傘) Umbrella	99 同(埋) With
		100 咁樣 Like this

Appendix 3. Parental concern question

Parents may have a range of concerns about their children’s speech or hearing, such as those listed on this showcard. Do you feel that [child’s name] has any of these issues?

1. no concerns

2. reluctant to speak

3. speech not clear to family

4. speech not clear to others

5. difficulty finding words

6. difficulty putting words together

7. doesn’t understand you when you speak

8. doesn’t understand others when they speak

9. voice sounds unusual

10. stutters, stammers, or lisps

11. other

Appendix 4. Percentile rankings for vocabulary scores on the NZ Mandarin monolingual and Cantonese monolingual CDI-sf.

Table

Percentile ranking	Mandarin N = 54	Cantonese^a N = 13
10th	12	11
20th	16	26
30th	26	34
40th	41	54
50th	58	67
60th	63	76
70th	78	84
80th	88	88
90th	96	95

Notes. ^aGiven the small sample size for the Cantonese sample, this percentile ranking should be used with caution.

Appendix 5. Percentile rankings for bilingual children’s vocabulary scores on the NZ Mandarin and Cantonese CDI-sf.

Table

Percentile ranking	Mandarin N = 92	Cantonese^a N = 25
10th	2	0
20th	6	2
30th	8	8
40th	12	12
50th	16	18
60th	19	24
70th	23	29
80th	33	46
90th	48	60

Notes. ^aGiven the small sample size for the Cantonese sample, this percentile ranking should be used with caution.

Appendix 6. Percentile rankings for multilingual children’s vocabulary scores on the NZ Mandarin and Cantonese CDI-sf.

Table

Percentile ranking	Mandarin N = 31	Cantonese^a N = 21
10th	0	0
20th	0	0
30th	1	2
40th	2	8
50th	4	12
60th	6	19
70th	12	27
80th	19	40
90th	46	52

Notes: ^aGiven the small sample size for the Cantonese sample, this percentile ranking should be used with caution.