Abstract
This is the first large-scale, quantitative study of the evaluative dimensions and potential predictors of Quebec-based parents’ attitudes towards childhood multilingualism. Such attitudes are assumed to constitute a determinant of parental language choices, and thereby influence children's multilingual development. The newly-developed Attitudes towards Childhood Multilingualism Questionnaire was used to gather data from 825 participants raising an infant/toddler aged 0–4 years with multiple languages in the home. The results revealed three separate dimensions: status and solidarity (the same dimensions found in attitudes towards individual languages) as well as cognitive development (not previously attested as a separate dimension). Participants’ approach to promoting multilingualism (specifically, whether they used the one-person-one-language-approach) and the combination of languages transmitted (specifically, whether this included a heritage language) correlated significantly with parental attitudes towards childhood multilingualism. Parents’ linguistic background and location within Quebec were not significant predictors of attitudes. The paper discusses implications and directions for further research.
Keywords
Many children around the world grow up exposed to multiple languages from an early age – be it because they are born into multilingual societies, because their parents have different native languages (L1s) from each other, or because they are descendants of migrants who have moved to societies in which languages other than their L1s are spoken. In the current context of globalization and high international mobility, it is likely that the number of children exposed to two or more languages from infancy will increase even further. Many parents are keen to raise proficiently multilingual children 1 (Byers-Heinlein & Lew-Williams, 2013), and research has shown that children are born with the capability to acquire more than one language without confusion or delay (e.g., Byers-Heinlein et al., 2010; Curtin et al., 2011). Yet, the language acquisition achievements of children with exposure to multiple languages are highly variable, and only some of these children will grow up to successfully and fluently use all of their languages (De Houwer, 2007). So what determines whether children exposed to multiple languages will become active, proficient multilinguals?
Numerous studies have demonstrated that multilingual development is shaped by the quantity and quality of language input (e.g., Marchman et al., 2017; Unsworth et al., 2019). Providing high-quality, balanced input is not a guarantee of subsequent active and fluent multilingualism – other factors also play a role, including children's language socialization through education as they grow older (De Houwer, 2007). Nevertheless, parental language choices in the early years are “an important […] determinant of children's future linguistic repertoires” (Caminal et al., 2018, p. 10). Research has shown that such language choices are, at least partly, determined by parents’ attitudes towards the specific languages involved in their children's upbringing (e.g., Kircher, 2019; Schüpbach, 2009). Less is known about the role of parental attitudes towards multilingualism as such, and towards childhood multilingualism in particular. Theoretical models have posited that attitudes towards childhood multilingualism do constitute an important factor in determining parental language choices, and thereby children's multilingual development (e.g., De Houwer, 1999, 2020; Pearson, 2007). Yet, there is a paucity of empirical research regarding this.
A more comprehensive understanding of parental attitudes towards childhood multilingualism is important for the advancement of theory, both with regard to multilingual development specifically and language attitudes more generally. Moreover, from a practical perspective, a better understanding of attitudes towards childhood multilingualism is fundamental to the development of appropriate measures to support families wishing to raise their children with multiple languages. Such support matters because children's multilingual development has micro- and macro-level implications. At an individual level, children's language outcomes can affect their well-being and healthy identity development as well as family dynamics and interpersonal relationships among family members (for an overview of the literature, see De Houwer, 2021). At a societal level, children who grow up to be active, proficient multilinguals and whose linguistic repertoires include minority languages can collectively contribute to the maintenance of these minority languages (King et al., 2008).
Here, we present a study aiming to do both: contribute to the advancement of theory and create knowledge that can inform measures to support multilingual families. The study investigates attitudes towards childhood multilingualism among parents in the Canadian province of Quebec. Quebec constitutes a complex and diverse linguistic environment. While both English and French have official status at the Canadian federal level, in Quebec, French is the sole official language whereas English is granted official minority language status by Canadian law. Yet English does not display the characteristics of a typical minority language: It is seen to hold high utilitarian value due to its international prestige and its role as the language of upward mobility in North America at large (e.g., Kircher, 2014, 2016). The province's public life is thus shaped by both the L1 French majority of the population and the L1 English minority (79.1% and 8.9%, respectively, including responses of individuals who have multiple L1s; Statistics Canada [StatCan], 2019). Regardless of their official status, both French and English effectively function as societal languages, especially in the province's urban center Montreal. Additionally, a growing percentage share of the province's population has L1s other than French and English (14.5%, including responses of individuals who have multiple L1s; StatCan, 2019). These L1 users of heritage languages – that is, “languages other than the dominant language (or languages) in a given social context” (Kelleher, 2010, p. 1) – include Indigenous peoples as well as immigrants and direct descendants of immigrants (2.3%, 13.7%, and 9.9% of the overall population, respectively; StatCan, 2016a). Many Quebecers have different L1 combinations, and while not all parents in the province transmit their L1s (Kircher, 2019), almost 17% of Quebec children aged 0–4 years are being raised in multilingual homes (Schott et al., 2021).
It is in this context that we investigated attitudes towards childhood multilingualism among parents who were raising an infant or toddler aged 0–4 years with more than one language in the home. Data were collected by means of the Attitudes towards Childhood Multilingualism Questionnaire (ACMQ), which we had developed for this purpose. Our aims were two-fold. Firstly, we aimed to investigate the nature of parents’ attitudes towards childhood multilingualism. Specifically, we set out to ascertain whether such attitudes constitute a unitary construct or whether they have different evaluative dimensions, as attitudes towards individual languages do (with status and solidarity being the main evaluative dimensions of attitudes towards individual languages; for further details, see below). Secondly, we aimed to determine variables that correlate with parental attitudes towards childhood multilingualism. The variables we investigated were: location within Quebec (i.e., whether participants lived in Montreal vs. in the rest of the province), linguistic background (i.e., whether participants had grown up monolingually vs. multilingually), approach to promoting multilingualism (i.e., the one-person-one-language approach vs. other approaches), and languages transmitted (i.e., whether a heritage language was included vs. no heritage language included in the combination of languages transmitted; for more details regarding these variables, see below). Our research questions (RQs) were:
This study is the first large-scale, quantitative investigation of the evaluative dimensions and potential predictors of Quebec-based parents’ attitudes towards childhood multilingualism. To contextualize it, we begin by providing an account of relevant aspects of language attitude theory as well as related empirical research.
Language Attitudes
Attitudes are composed of three components: affect – the feelings elicited by an attitude object, cognition – the beliefs held about the attitude object, and conation – the behavior directed at the attitude object (Bohner, 2001). The relationship between attitudes and behavior is not a simple, direct one. Individuals’ behavior depends not only on their attitudes but also on a range of other variables, including the specific context and the immediate consequences that the behavior can be expected to have (Banaji & Heiphetz, 2010). Hence, rather than being equated with behavior, attitudes are considered to be dispositions to behave in certain ways. They influence (rather than fully determine) actions (Bohner, 2001).
Previous studies have shown that attitudes towards individual languages influence a wide range of linguistic behaviors, including what languages people decide to learn (e.g., Gardner & MacIntyre, 1993), the frequency with which they use their languages (e.g., Edwards & Fuchs, 2018), and whether they transmit these languages to their children (e.g., Kircher, 2019; for a more detailed overview of the consequences of language attitudes, see Dragojevic et al., 2021).
Language attitudes are traditionally defined as “any affective, cognitive or [conative] index of evaluative reactions towards different varieties and their speakers” (Ryan et al., 1982, p. 7) – or, more inclusively, their users. While this classic definition only makes reference to attitudes towards different varieties (i.e., languages, dialects, accents), attitudes towards particular linguistic phenomena such as vocal fry (e.g., Yuasa, 2010) and code-switching (e.g., Dewaele & Wei, 2014) also fall under the remit of language attitudes – and so do attitudes towards multilingualism.
Attitudes towards multilingualism are conceptually distinct from attitudes towards the specific languages involved in a multilingual context (Baker, 1992). However, there is little work on attitudes towards multilingualism itself – and while insightful, the work that does exist is mostly of a qualitative and context-specific nature that makes it difficult to generalize (e.g., Jaworska & Themistocleous, 2017; Schüpbach, 2009). There is even less previous research focusing specifically on attitudes towards childhood multilingualism (for exceptions, see Baker, 1992; and Surrain & Luk, 2019, which we discuss below). In the following sections, we state our hypotheses regarding the dimensionality and predictors of such attitudes, and we provide an account of the literature that motivated these hypotheses.
The Dimensionality of Attitudes
Attitudes towards individual languages are generally considered to have two main evaluative dimensions: status and solidarity. A language with high status is associated with power, economic opportunity, and upward social mobility. Attitudes on the status dimension are thus linked to utilitarian value (Gardner & Lambert, 1972). By contrast, a language that is evaluated positively on the solidarity dimension elicits feelings of attachment and belonging – it holds “vital social meaning and […] represent[s] the social group with which one identifies” (Ryan et al., 1982, p. 9). Attitudes on the solidarity dimension are thus linked with social identity and ingroup loyalty. Studies conducted over several decades and in numerous parts of the world (for an overview, see Giles & Watson, 2013) have yielded empirical support for the notion that status and solidarity are independent evaluative dimensions with “a universal importance for the understanding of attitudes” (Ryan et al., 1988, p. 1073; see also Dragojevic & Giles, 2016).
Some studies have found there to be more than two dimensions, and some have labeled them differently. For instance, Lambert (1967) distinguished between one status-type factor, termed ‘competence’, and two solidarity-type factors, labeled ‘personal integrity’ and ‘social attractiveness’. Cargile (1997), on the other hand, distinguished between one solidarity-type factor, termed ‘attractiveness’, and two status-type factors, labeled ‘status/dynamism’ and ‘job suitability’. The dimensions of status and solidarity should thus not be seen as exhaustive. Nevertheless, factor analytic research has generally shown status-type factors and solidarity-type factors to be distinct from one another. 2
As noted above, there is little quantitative research regarding attitudes towards multilingualism. Baker’s (1992) and Surrain and Luk’s (2019) work are notable exceptions. Baker (1992) studied Welsh-English multilingualism in Wales via a questionnaire distributed to 797 adolescents; Surrain and Luk (2019) investigated Spanish-English multilingualism in the U.S. via a questionnaire distributed to 422 adults. Neither found attitudes towards multilingualism to have separate dimensions. Yet, it should be noted that neither included items that were explicitly phrased to tap status and solidarity as distinct dimensions. Moreover, some of their wording was context-specific, making reference to the languages involved in the specific research contexts rather than to multilingualism as such.
Surrain and Luk’s (2019) scale of perceptions of childhood multilingualism in the U.S. – which they distributed to 321 parents as part of their aforementioned questionnaire – constitutes the only known research instrument that focuses on eliciting quantitative data regarding attitudes towards multilingualism in children. This scale comprizes two items regarding the importance of speaking, reading, and writing multiple languages; one item each regarding the potential positive and negative cognitive consequences of acquiring multiple languages simultaneously; one item each regarding the future success in education and employment that might result from acquiring multiple languages simultaneously; one item regarding the facilitation of understanding people from different backgrounds; as well as one item about the importance of English as the only language that children need to speak well. All items loaded onto one factor, so – like attitudes towards Spanish-English multilingualism more generally – the researchers found perceptions of childhood multilingualism in the U.S. to be a unidimensional construct. However, again, the researchers had not included items that were explicitly designed to tap status and solidarity, and again, there was some context-specific wording. Moreover, they did find items concerning parental beliefs about children's developmental (i.e., cognitive) trajectories to be only weakly correlated with the other items. The researchers thus discuss the possibility that parental beliefs regarding children's developmental trajectories may constitute “another dimension that requires a separate subscale” (Surrain & Luk, 2019, p. 33).
In the first phase of our research project about Quebec-based parents raising children multilingually, we conducted exploratory interviews and focus groups with 27 Montreal-based parents who were raising infants/toddlers aged 0–3 years with multiple languages in the home (Ballinger et al., 2020). The resulting qualitative data indicate that these parents gave much thought to the potential impact of multilingualism on their children's cognitive development, with some participants raising concerns while others anticipated cognitive advantages. Additionally, parents discussed points relating, firstly, to the utilitarian value of childhood multilingualism, and secondly, to the perceived connection between childhood multilingualism and ingroup ties. These findings echo Surrain's results from a previous study of qualitative data, collected in interviews with 14 Spanish-speaking mothers in the U.S. (Surrain, 2021). These mothers deemed Spanish-English childhood multilingualism to be, firstly, “essential for their child's future economic prospects”, and secondly, crucial for their child's “connections with family and cultural heritage” (Surrain, 2021, p. 1168).
The existing literature thus suggests that parents distinguish between status-related and solidarity-related aspects of childhood multilingualism (Ballinger et al., 2020; Surrain, 2021) as well as aspects related to children's cognitive development (Ballinger et al., 2020; Surrain & Luk, 2019). In light of this, and taking account of the aforementioned dimensionality of attitudes towards specific languages, our hypotheses (Hs) were:
Potential Predictors of Attitudes
The variables we investigated as potential predictors of attitudes towards childhood multilingualism were: location within Quebec, linguistic background, approach to promoting multilingualism, and languages transmitted.
Location within Quebec. Quebec's urban center Montreal is even more linguistically diverse than the rest of the province. More than 21% of residents speak at least three languages, making it Canada's most multilingual city (StatCan, 2016b; see also Schott et al., 2021). Most of Quebec's L1 English speakers live in Montreal, and while Indigenous peoples make up less than 1% of the city's population, over 23% of Montrealers are foreign-born (StatCan, 2016a). Notably, around 30% of children in the city have at least one first-generation immigrant parent (Institut de la statistique du Québec, 2018) and the occurrence of infants/toddlers aged 0–4 years who are being raised with multiple languages in the home is thus particularly high (24.1%, compared to 16.7% in the province overall; Schott et al., 2021). Previous research has attested attitudinal differences regarding French and English between inhabitants of the city of Montreal, its suburbs, and the rest of the province (Kircher, in preparation). No previous research exists regarding the relationship between location and attitudes towards childhood multilingualism.
Linguistic background. Just like having a particular L1 has been found to affect individuals’ attitudes towards that specific language (e.g., Genesee & Holobow, 1989; Kircher, 2014; Lambert et al., 1960), being multilingual has been found to affect individuals’ attitudes towards multilingualism. Surrain and Luk (2019) showed that, in the U.S., multilingual parents perceived Spanish-English childhood multilingualism more favorably than monolingual parents did. This was echoed in the findings from the first phase of our research project: Montreal-based parents mentioned that people “who had little or no experience with simultaneous bilingualism” were often critical of the participants’ decision to raise their children multilingually – but those who had grown up with multiple languages themselves dismissed such concerns, making reference to their own successful experiences with multilingualism (Ballinger et al., 2020, pp. 6–7). It remains to be ascertained whether the relationship between parents’ own linguistic background and their attitudes towards childhood multilingualism is a systematic one.
Approach to promoting multilingualism. There are different strategies parents can employ to ensure their children receive approximately equal exposure to their languages in order to promote multilingual development. It is common for each parent to use multiple languages with the child, or for one parent to use multiple languages while the other only uses one (De Houwer, 2021). In heritage language (HL) families, parents often use only the HL in the home, and they employ the societal language(s) outside their own four walls (Ballinger et al., 2020). Nevertheless, the so-called ‘one-person-one-language’ approach is the most “(in)famous” strategy (De Houwer, 2021, p. 8). As Byers-Heinlein and Lew-Williams (2013) explain, what matters is the quality and quantity of the input, not the approach used to provide it. Nevertheless, the one-person-one-language approach has much prominence in debates about raising multilingual children. No research exists regarding the relationship between parents’ attitudes towards childhood multilingualism and their strategy for promoting it.
Languages transmitted. The qualitative data from the first phase of our research project (Ballinger et al. 2020) revealed clear differences between parents who were raising their children with French and English (non-HL parents) compared to parents who were transmitting one or more HLs in addition to French and/or English (HL parents). The non-HL parents expressed almost unequivocal optimism regarding their children's multilingual development. By contrast, the HL parents were much less optimistic. The differences were so pronounced that they motivated our second hypothesis:
HLs themselves do not tend to hold much utilitarian value in the Quebec context. Neither Indigenous nor immigrant HLs hold official status; they do not receive the same institutional support as French and English; and there is little formal support for their intergenerational transmission. Knowledge of HLs is not a requirement in education or for career advancement. However, multilingualism which includes French and/or English opens many doors in the province. The participant sample of Ballinger et al. (2020) did not include any parents who were solely transmitting HLs, and in fact, previous research suggests that it is rather rare for HL speakers in Quebec to raise their children without French and/or English (Kircher, 2019). Consequently, we hypothesized as follows:
Raising a child with languages other than French and/or English in Quebec requires extra effort. For instance, parents need to ensure sufficient input and obtain resources such as children's books in the HLs, which are often not readily available (Ahooja et al., under review). Several HL parents in Ballinger et al. (2020) felt that it was difficult to be the only one responsible for passing on their HL. They noted that the transmission of HLs is different from that of French and English, to which children can easily receive exposure not only from their parents but also outside the home, and where there is a prospect of future language socialization through education. Going to great lengths to raise children with HLs is likely to be the result of a strong conviction that acquiring these HLs entails a fundamental benefit for the children – a benefit that is linked to the importance of the HLs to the social group(s) one identifies with; because knowledge of the HLs enables communication and connections with the other HL speakers in the family and the community, because knowledge of the HLs strengthens ingroup ties, and because it allows children to access their cultural heritage. (This goes beyond the utilitarian value mentioned above, which relates solely to socio-economic advancement.) Parents who lack this conviction will probably not transmit their HLs at all, or at least not beyond a symbolic level. Since our participant sample only included parents who were raising their children multilingually, we thus hypothesized:
Children's cognitive development was at the forefront of many parents’ minds in Ballinger et al. (2020), with HL parents expressing more worry about potentially negative consequences of childhood multilingualism than non-HL parents. Amongst other things, the HL parents voiced more worries about their children mixing languages, about overloading their children with too many languages at once, and about multilingualism causing delays to their children's language development. Therefore, we hypothesized as follows:
Methodology
Method
The data for this study were collected by means of the ACMQ, which, prior to the main data collection, was piloted amongst 24 Quebec-based parents to test the items and instructions. The ACMQ is part of a larger questionnaire by means of which we also investigated Quebec-based parents’ concerns about raising children multilingually (Quirk et al., in preparation), their language practices (Phillips et al., in preparation), and their use of resources to promote multilingual development (Ahooja et al., under review). Here, we focus only on those aspects of the project that pertain to parents’ attitudes. Our hypotheses and data analysis plan were pre-registered prior to examination of the data. This pre-registration, the ACMQ, the raw data, and the R analysis code are available at: https://osf.io/qcvse/.
Attitudes towards childhood multilingualism. Attitudes were elicited by means of twelve items whose response options were five-point Likert scales (with 1 meaning don’t agree at all, 5 meaning agree completely). There were four items each that were designed to tap status (29, 31, 34, and 36), solidarity (30, 33, 35, and 38), and cognitive development (27, 28, 32, and 37; see Table 2) – that is, our hypothesized dimensions. In the absence of existing (non-context-specific, non-language-specific) scales for the elicitation of attitudes towards childhood multilingualism, we developed the ACMQ items based on the findings from Ballinger et al. (2020) and existing knowledge regarding the dimensionality of attitudes towards individual languages. Three of our items (29, 34, and 37) tapped similar points as Surrain and Luk’s (2019) items developed for the U.S. context. The other nine were different since we set out to tap status, solidarity, and cognitive development – and we included several items to tap each hypothesized dimension (see above) to ensure validity. The ACMQ items purposely did not make explicit reference to either the research context or to specific languages.
Location within Quebec. An open-ended item asked participants where they lived. In line with the findings of the aforementioned study by Kircher (in preparation), the data were initially classified into three categories: participants living (1) on the island of Montreal, (2) in the suburbs of Montreal, and (3) in the rest of Quebec. As the attitudinal differences between the first two categories turned out not to be significant, their data were combined. The resulting categories used for the final analyses were: participants living (1) in Montreal, and (2) in the rest of Quebec. 3
Linguistic background. An open-ended item asked participants what language(s) they considered to be their mother tongue(s), specifying that this meant the language(s) they had learned at home, as a child, and that they still understood. This was based on StatCan's (2009) definition of L1 as “the first language learned at home in childhood and still understood”, acknowledging that individuals can have more than one L1. The participants were classified as having grown up (1) with only one L1, and (2) with multiple L1s.
Approach to promoting multilingualism. Five-point Likert scales (with 1 meaning never, 5 meaning all the time) were used to elicit the frequency with which participants used French, English, and other languages with their child. Additionally, a multiple-choice item was included to find out what language(s) the child's other caregiver(s) – if applicable – used with the child. Participants were categorized as using (1) the one-person-one-language approach, and (2) any other approach. We applied the one-person-one-language category not only for participants who followed this approach very strictly (reporting use of one language with their child as 5, and 1 for other languages) but also for those who followed it slightly more flexibly. This decision was based on the fact that, in a multilingual environment like Quebec, even parents who have adopted the one-person-one-language approach may occasionally employ another language in situations where they and their child are in the presence of someone who does not understand their habitual language. Participants who did this only very rarely (reporting use of a non-habitual language as 2 on the scale) were thus also included in the one-person-one-language category.
Languages transmitted. The languages transmitted to a participant's child were determined based on the same five-point Likert scales used to elicit the frequency with which each participant used French, English, and other languages with their child (with 1 meaning never, 5 meaning all the time). If a parent reported using a language with their child at a frequency greater than 2, this was coded as the language being transmitted. Additionally, the aforementioned multiple-choice item was used to establish what language(s) were being transmitted by a second caregiver – if applicable. Based on these data, participants were classified into two categories: those whose children were being raised (1) with at least one HL, and (2) without any HLs. 4
Procedure
The ACMQ was distributed online via the mailing lists of the Concordia Infant Research Lab and the McGill Infant Speech Perception Lab, which include Quebec-based parents largely recruited from government birth lists who have agreed to be contacted for research purposes. In order to further encourage parents to participate, we also shared the ACMQ via other mailing lists as well as traditional and social media, including targeted ads for the more remote regions. The ACMQ was available in both English and French, and participants could respond in the language of their choice. Participants were also encouraged to share the questionnaire with further potential participants (snowball sampling). Data collection took place over a period of six months, from September 2020 until February 2021.
Participants
The participant sample consists of 825 parents who, at the time of data collection, lived in Quebec, were over the age of 18, and were raising a typically-developing infant or toddler aged 0–4 years with multiple languages in the home. 5 72.0% of the participants lived in Montreal and 28.0% lived elsewhere in Quebec. Their ages ranged from 23 to 56, with a mean age of 35 years. 92.9% of participants identified as female, 6.4% identified as male, 0.5% indicated a gender other than female or male, and 0.2% did not provide information concerning their gender. 77.5% had grown up monolingually while the remaining 22.5% had been raised multilingually themselves. 38.3% had French as their L1/as one of their L1s, 34.5% had English as their L1/as one of their L1s, and 46.4% had a language other than French or English as their L1/as one of their L1s. As for the child-rearing constellations: 2.3% were raising their children alone, 1.8% with partner(s)/parent(s) living in a different household, 95.4% with partner(s)/parent(s) living in the same household, and 0.1% were raising their children in other constellations. Only 11.6% were using the one-person-one-language approach to promote their children's multilingualism while 88.4% used different approaches. 60.2% of the participants were transmitting HLs to their children (HL parents) while 39.8% were passing on English and French only (non-HL parents). The specific language combinations with which the participants’ children were being raised are shown in Figure 1 (in the supplementary materials folder on OSF). The children's ages ranged from 3 weeks to 48 months, with a mean age of 23.5 months. 49.8% of participants described their children's sex as female, 50.1% as male, and 0.1% indicated that their child had a sex other than female or male.
Statistical Analyses
As noted above, the first research aim was to determine whether parents’ attitudes towards childhood multilingualism constitute a unitary construct or whether they have different evaluative dimensions. Conway and Huffcott (2003, p. 150) state that “[i]f a researcher's purpose is to understand the [underlying] structure of a set of variables […], then use of a common factor model such as […] maximum likelihood factoring represents a high-quality decision”. Therefore, exploratory factor analysis with maximum likelihood factoring was performed. An oblique rotation method (direct oblimin) was chosen since “most factors related to human cognition […] can be assumed to be related in some way” (Plonsky & Gonulal, 2015, p. 22). Subsequently, multiple regressions were performed to meet our second research aim – that is, to ascertain which of the aforementioned variables correlated with parental attitudes towards childhood multilingualism, thus constituting potential predictors.
Results
The Dimensionality of Attitudes
The Kaiser-Meyer-Olkin index (.83) and Bartlett's Test of Sphericity (sig. <.001) indicated that the data were suited for factor analysis. Parallel analysis revealed three components, and the factor correlation matrix (see Table 1) demonstrated that the three extracted factors were indeed interrelated and that the obliquely rotated solution was thus adequate.
Factor Correlation Matrix.
The factor matrix (see Table 2) showed that the factor loadings support a division into our three hypothesized dimensions of attitudes towards childhood multilingualism – that is, status, solidarity, and cognitive development. Items 27, 28, 29, 32, and 37 loaded onto factor 1, 6 indicating that they all tapped the same dimension – namely cognitive development; items 31, 34, and 36 loaded onto factor 2, indicating that they all tapped the same dimension – namely status; and items 30, 33, 35, and 38 loaded onto factor 3, indicating that they all tapped the same dimension – namely solidarity. The items that loaded onto each factor had high internal consistency: α = .81 for the cognitive development factor, α = .78 for the status factor, and α = .67 for the solidarity factor. Together, these three factors accounted for much of the variance: cognitive development for 22.6%, status for 19.0%, and solidarity for 16.8% – together 58.4%. Notably, the three factors we found were correlated: Participants who reported more positive attitudes on one dimension were also more likely to report positive attitudes on the other dimensions.
Factor Matrix.
All items loaded onto the factor pertaining to the dimension they were designed to tap (see above) – with the exception of item 29, Knowing more than one language will, in the long term, help my child do well in school. This item had been designed to tap attitudes on the status dimension but, in fact, it loaded onto the cognitive development factor. This was likely due to the (unintentionally ambiguous) phrasing of the item. It had been designed to refer to ‘doing well’ in the sense of achieving good grades, which allow for subsequent economic opportunities and social mobility (i.e., status). However, it can also be interpreted as referring to ‘doing well’ in the sense of learning easily and understanding subject matters (i.e., cognitive development). It appears that participants mainly interpreted the item to mean the latter. The item could be reworded in future iterations of the ACMQ, in a way that unequivocally pertains to attitudes on the status dimension. Yet, this has no bearing on the overall results of the factor analysis.
Potential Predictors of Attitudes
Following the factor analysis, the items that loaded onto each factor were combined and their mean was calculated to create an overall value for attitudes on the status dimension, an overall value for attitudes on the solidarity dimension, and an overall value for attitudes on the cognitive development dimension. 7 Attitudes on all three dimensions were positive, with an overall mean of 4.74 (SD = .51) for status, 4.55 (SD = .62) for solidarity, and 4.43 (SD = .65) for cognitive development (recall that 5 was agree completely for all attitude items). 8 Multiple regressions were then performed to investigate which of the aforementioned variables correlated with parental attitudes towards childhood multilingualism on each dimension, thus constituting potential predictors. The reference levels were: inside of Montreal for location within Quebec, monolingual for linguistic background, an approach other than one-person-one-language for approach to promoting multilingualism, and no HL included for languages transmitted.
The initial regression results showed that two variables had a significant association with attitudes on the status dimension (see Table 3): parents’ approach to promoting multilingualism, with the adoption of the one-person-one-language approach being associated with less positive status-related attitudes (see Figure 2 in the supplementary materials folder on OSF); and the languages transmitted, with the presence of a HL also being associated with less positive status-related attitudes (see Figure 3 in the supplementary materials folder on OSF). For attitudes on the solidarity dimension, the languages transmitted constituted the only variable that had a significant association (see Table 4) – this time with the presence of a HL being associated with more positive solidarity-related attitudes (see Figure 4 in the supplementary materials folder on OSF). For attitudes on the cognitive development dimension, no significant effects were found (see Table 5).
Multiple Regressions: Coefficients for All Variables (Full Model) and Statistically Significant Predictors Only (Pruned Model) – Status Dimension.
Notes: Full model R2 = .03, Adjusted R2 = .02, F(4,815) = 5.31, p < .001, Pruned model R2 = .02, Adjusted R2 = .02, F(2,817) = 9.51, p < .001.
Multiple Regressions: Coefficients for All Variables (Full Model) and Statistically Significant Predictors Only (Pruned Model) – Solidarity Dimension.
Notes: Full model R2 = .02, Adjusted R2 = .02, F(4,815) = 4.21, p < .01, Pruned model R2 = .02, Adjusted R2 = .02, F(1,818) = 16.28, p < .001.
Multiple Regressions: Coefficients for All Variables – Cognitive Development Dimension.
Notes: R2 = .00, Adjusted R2 = -.00, F(4, 813) = .87, ns.
Notably, the correlation matrix for all variables showed that parents in Montreal were more likely to transmit a HL than those elsewhere in the province; multilingual parents were more likely to follow the one-person-one-language approach than monolingual parents; and the one-person-one-language approach was more common amongst non-HL parents than amongst HL parents (see Table 6). However, when additional exploratory models were run for each of the three dimensions including interaction terms, none of these were significant, and thus they were not included in the models.
Correlations Matrix for All Variables.
As planned in the pre-registration, we then repeated the regressions with only those variables that had been found to be statistically significant in the initial rounds of analysis – that is, approach to promoting multilingualism (for attitudes on the status dimension) and languages transmitted (for attitudes on the status and the solidarity dimension). In these pruned models, the same variables remained significant. 9 However, both full and pruned models accounted for only a very small share of the variance: 2% for each dimension (see Tables 3 and 4).
Discussion
The Dimensionality of Attitudes
RQ1 asked: Do parents’ attitudes towards childhood multilingualism constitute a unitary attitude factor or do they have different evaluative dimensions? The factor analysis results revealed three distinct factors, thus providing support for H1: The participants’ attitudes did indeed have different evaluative dimensions. Our study is thus the first to confirm the multi-dimensionality of attitudes towards childhood multilingualism. As hypothesized, two of the factors we found corresponded to the evaluative dimensions present in attitudes towards individual languages, namely status and solidarity (H1a). Our findings therefore show that the “universal importance” that status and solidarity have for the understanding of attitudes (Ryan et al., 1988, p. 1073) applies not only to attitudes towards individual languages – it also extends to attitudes towards childhood multilingualism. Moreover, as hypothesized, the third factor we found corresponded to an evaluative dimension pertaining to children's cognitive development (H1b). Attitudes on the cognitive development dimension manifested in parents’ belief that knowing multiple languages entails cognitive benefits for their child. Specifically, the ACMQ revealed that our participants believed childhood multilingualism would make their children better learners and more flexible thinkers as well as improving their reasoning abilities and facilitating the acquisition of further languages later in life. Our study is the first to attest cognitive development as a separate dimension of attitudes towards childhood multilingualism.
As noted above, the few previous factor analytic studies of attitudes towards (childhood) multilingualism – namely Baker (1992) and Surrain and Luk (2019) – had found a single latent construct. Yet, Surrain and Luk's (2019) factor loadings did lead them to speculate about beliefs regarding children's developmental trajectories being a separate dimension. Unlike Baker (1992) and Surrain and Luk (2019), we employed items that were explicitly phrased to tap status, solidarity, and cognitive development as distinct dimensions – and we employed several items to tap each dimension, to ensure validity. Moreover, unlike these previous studies, we avoided phrasing our items in a context-specific manner (i.e., no reference was made to Quebec or Canada), and we made reference to multilingualism as such rather than to the specific languages involved in the research context (i.e., English, French, and HLs). It seems likely that the design of the ACMQ items is what allowed us to provide evidence for status, solidarity, and cognitive development as separate evaluative dimensions of parental attitudes towards childhood multilingualism.
The finding that the three factors were correlated could be due to the fact that they all relate to human cognition and are thus conceptually linked (cf. Plonsky & Gonulal, 2015) – or it could indicate that there is one underlying macro-factor for attitudes towards childhood multilingualism, which has several sub-factors. Further research is necessary to ascertain this. Notably, together, the three factors we found accounted for much of the variance in our data: 58.4%. This indicates the importance of these dimensions. However, the fact that they did not account for all of the variance suggests that further dimensions of parental attitudes towards childhood multilingualism may exist. More research is needed to investigate this. Nonetheless, our study makes an important contribution to attitude theory by advancing knowledge regarding the dimensionality of attitudes towards childhood multilingualism.
Potential Predictors of Attitudes
RQ2 asked: What demographic and socio-linguistic variables correlate with parents’ attitudes towards childhood multilingualism, thus constituting potential predictors? The regression results lent support to H2: The combination of languages transmitted – and specifically, the inclusion of a HL – constituted the main predictor of participants’ attitudes towards childhood multilingualism. This is in line with the findings from the first phase of our research project (Ballinger et al., 2020). However, for two of the dimensions, namely status and cognitive development, the combination of languages transmitted did not correlate with attitudes in the manner we had hypothesized.
Regarding the status dimension, we had hypothesized that HL parents and non-HL parents would hold equally positive attitudes towards childhood multilingualism (H2a). Yet, the regression results revealed a significant difference: The presence of a HL was associated with less positive status-related attitudes. As noted above, the items in the ACMQ were phrased in a non-context-specific manner, and without reference to specific languages. Nevertheless, it is possible that the participants interpreted the status items as pertaining to the Quebec context, and as referring to multilingualism involving the specific languages with which their children were being raised. This would explain why the non-HL parents (whose children were growing up with French and English – i.e., two languages that hold high utilitarian value in Quebec society) expressed more positive attitudes towards childhood multilingualism than the HL parents (whose children were growing up with at least one HL, i.e., a language that holds little utilitarian value in Quebec). However, further research is necessary to confirm the underlying reason for the status-related results.
Regarding the solidarity dimension, we had hypothesized that HL parents would hold more positive attitudes towards childhood multilingualism than non-HL-parents (H2b). This was borne out by the regression results: The presence of a HL was indeed associated with more positive solidarity-related attitudes. As explained above, raising children with languages other than the main societal language(s) of a community requires extra effort and is not always easy. We thus assumed that parents who transmit their HLs despite the added effort and difficulty hold a strong conviction that acquiring these HLs will entail a fundamental benefit for their children – a benefit linked to the importance of the HLs to the social group(s) they identify with. This serves as a likely explanation for the fact that the HL parents in this study did hold more positive solidarity-related attitudes towards childhood multilingualism than the non-HL parents.
Regarding the cognitive development dimension, we had hypothesized that non-HL parents would hold more positive attitudes towards childhood multilingualism than HL-parents (H2c). However, this was not borne out by our data. In fact, the presence of a HL did not covary with attitudes on the cognitive development dimension in a significant way at all. H2c had been based on the HL parents’ more vocal expression of their worries about potentially negative cognitive consequences of childhood multilingualism in the first phase of our project (Ballinger et al., 2020). Yet, the regression results indicated that this does not constitute a systematic and significant trend. It seems that the participants were, after all, at least implicitly aware that the specific languages with which children are raised do not entail differences to the cognitive processes involved in growing up multilingually.
It is unclear why the adoption of the one-person-one-language approach was associated with less positive attitudes towards childhood multilingualism on the status dimension. One possibility is that proficiency in the languages that are transmitted may have affected parental attitudes towards childhood multilingualism, and that those participants who had adopted the one-person-one-language approach were less multilingual themselves. This remains to be ascertained. Moreover, for both the status and the solidarity dimension, the variables we found to correlate with attitudes in a significant manner – that is, approach to promoting multilingualism (for status) and languages transmitted (for status and solidarity) – accounted for only a small share of the variance: 2% for each dimension. There was also a difference regarding the amount of variation in participants’ responses on the status and the solidarity dimension. More research is needed to investigate what is at the root of this differences in variation, and what other variables constitute further predictors of parental attitudes towards childhood multilingualism.
Nonetheless, by providing an initial understanding of such predictors, our study makes an important contribution: It lays the groundwork for the development of appropriate measures to support families who wish to raise their children multilingually. Knowing which groups of parents hold less positive attitudes towards childhood multilingualism, and on which dimensions, will facilitate the tailoring of support measures specifically to those groups. As noted above, such support matters because children's multilingual development not only has macro-level implications for the maintenance of minority languages (King et al., 2008) but also crucial micro-level implications for children's well-being and healthy identity development, as well as family dynamics and interpersonal relationships among family members (see De Houwer, 2021).
Limitations
Certainly, there are limitations to the study presented here. For instance, the aim of the ACMQ was clearly recognizable, which might have caused responses to be affected by social desirability bias (Baker, 1992). However, due to the anonymity offered by online questionnaires, the likelihood of this is comparatively low (Dewaele, 2018). Had data been gathered by means of other direct methods, such as face-to-face interviews or focus groups, the risk of social desirability bias would certainly have been higher (Oppenheim, 2000). Indirect, experimental methods to elicit attitudes towards multilingualism as a linguistic phenomenon (rather than attitudes towards specific languages) are yet to be developed. Consequently, despite its inherent limitations, a questionnaire constituted the most appropriate choice of method for this study.
A further limitation is that we only collected data from parents who had already made the decision to raise their children multilingually – because their attitudes are likely to be different from those of parents raising their children monolingually. Moreover, while we reached out to Quebec-based parents from the government birth lists, and we made use of targeted campaigns (and snowball sampling) to further encourage participation among parents especially in the more remote regions of the province, our participant sample is not representative. We thus make no claims regarding the generalization of our findings to the Quebec population at large.
Finally, the correlational nature of the study needs to be acknowledged: The findings allow us to conclude which predictor is related to parental attitudes towards childhood multilingualism, but not whether this relationship is in fact causal – even though causality does seem probable, at least with regard to this particular participant sample of parents raising their children multilingually. Notwithstanding these limitations, the study's findings provide meaningful insights into the attitudes towards childhood multilingualism held by Quebec-based parents raising infants/toddlers aged 0–4 years with multiple languages in the home. Should these findings hold true, they would have important implications for both theory and practice.
Directions for Further Research
The participants in this study, who were all raising their children with more than one language, held very positive attitudes towards childhood multilingualism. Their attitudes were thus consistent with their behavior. Based on the findings of this study, the next step of our work will be to ascertain whether, and to what extent, attitudes towards childhood multilingualism – on each of the three dimensions we found – in fact determine parental language choices (Phillips et al., in preparation). As noted above, theoretical models have long posited that attitudes towards childhood multilingualism constitute an important determinant of parental language choices, and thereby children's multilingual development (e.g., De Houwer, 1999, 2020; Pearson, 2007). Yet, to date, there is little empirical research directly testing this hypothesis. The knowledge of the dimensionality of attitudes towards childhood multilingualism that is provided by the present study will allow for a much more nuanced analysis of the relationship between parents’ language attitudes and their language choices in future research. Our findings will therefore also contribute to models of intergenerational language transmission.
Subsequently, more research is necessary to further investigate the dimensionality of attitudes towards childhood multilingualism as well as additional predictors. In addition to focusing on the parents of infants/toddlers who are growing up in multilingual homes, future research should also consider caregivers of infants/toddlers whose primary exposure to their LXs takes place in contexts outside the home, such as daycare. Moreover, moving forward, research building on the study presented here should extend to parents raising children older than 0–4 years as well. Previous studies have found parents’ language practices in the home to shift in response to changing life circumstances, such as altered family compositions and children's exposure to new language environments in daycare or school (Prevoo et al., 2011; Schecter et al., 1996; see also Surrain, 2021). Longitudinal research and studies with parents raising children from a range of age groups will show whether these shifting language practices are linked with changing attitudes, thereby providing insights into the trajectories of parental attitudes towards childhood multilingualism.
Conclusion
By revealing parental attitudes towards childhood multilingualism to have distinct evaluative dimensions, the research presented here makes an important contribution to attitude theory. Moreover, the knowledge derived from this study will allow for a more nuanced analysis of the relationship between parents’ language attitudes and their language choices in future research, thereby also contributing to models of intergenerational language transmission. By providing an initial understanding of the predictors of parental attitudes towards childhood multilingualism, the present study lays the groundwork for the development of appropriate measures to support families wishing to raise their children with multiple languages. Future research will undoubtedly further refine our understanding of this topic, thereby continuing the advancement of theory as well as facilitating the creation of support measures tailored to those multilingual families who need them most.
Supplemental Material
sj-docx-1-jls-10.1177_0261927X221078853 - Supplemental material for Quebec-based Parents’ Attitudes Towards Childhood Multilingualism: Evaluative Dimensions and Potential Predictors
Supplemental material, sj-docx-1-jls-10.1177_0261927X221078853 for Quebec-based Parents’ Attitudes Towards Childhood Multilingualism: Evaluative Dimensions and Potential Predictors by Ruth Kircher, Erin Quirk, Melanie Brouillard, Alexa Ahooja, Susan Ballinger, Linda Polka and Krista Byers-Heinlein in Journal of Language and Social Psychology
Footnotes
Acknowledgments
We are grateful to Hilary Killam and Ethan Kutlu for their support with the R code, and to Howie Giles and the anonymous JLSP reviewers for their constructive feedback.
Funding
This work was supported by a Research Incubator Award from the Centre for Research on Brain, Language and Music (CRBLM), an Insight Grant to KBH from the Social Sciences and Humanities Research Council of Canada [435-2019-1032], and a grant to KBH from the National Institutes of Health [1R01HD095912-01A1]. KBH holds the Concordia University Research Chair in Bilingualism and Open Science, which also supported this research.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
