In her very interesting Keynote Article, Johanne Paradis gives a clear overview of recent research at the interface of bilingual development and child language disorders, and highlights its theoretical and clinical implications. She raises the challenging question of “whether bilingualism can be viewed as a kind of ‘therapy’ for SLI.” At first sight, this is perhaps a surprising question, because one of the predominant views in the literature is that bilingual children with specific language impairment (SLI) will exhibit difficulties and perhaps a “double delay.” It is this challenging question that we consider in more detail here.
More specifically, we would like to take a closer look at what Paradis' perspective tells us about the opportunities and vulnerabilities in early child bilingualism in general, for if bilingualism can be advantageous for children with SLI, it follows that it can also be advantageous for typically developing (TD) bilingual children. Paradis, who is one of the few linguists performing research on the interface of bilingual development and language disorders, systematically stresses the positive aspects and the advantages of bilingualism in different child populations. She also rightly reminds us that children who learn two languages before puberty are the majority worldwide.
Paradis furthermore stresses the importance of comparing different bilingual populations to learn more about the possible interaction between factors of success and failure in bilingual development. On the basis of her discussion, we can distinguish four domains in which bilingual child populations may generally differ: (a) age of onset, (b) societal contexts, (c) cognitive maturity, and (d) languages involved.
In the following we will attempt to disentangle the different factors of success and failure mentioned by Paradis in relation to the development of bilingual children with SLI, and try to establish what they tell us more generally about early child bilingualism in different populations.
AGE OF ONSET
In child bilingualism studies, differences in exposure patterns are related to well-known factors such as the age of onset of acquisition, input quantity, length of exposure, and language dominance. In these studies (e.g., Meisel, Reference Meisel2007; Montrul, Reference Montrul2008; Paradis, Reference Paradis, Hoff and Shatz2007), a distinction is often made between children growing up bilingually from birth, who are simultaneously bilingual (two first languages [2L1]) children, and sequential bilingual children who start acquiring their second language later (cL2), say, after age 3 (e.g., MacLaughlin, Reference McLaughlin1978). Paradis convincingly argues that such a distinction based on the age of onset can never be the only factor to take into account when explaining differences in bilingual language development. Cornips and Hulk (Reference Cornips and Hulk2008) make a similar point in a recent article on the acquisition of grammatical gender in Dutch in different bilingual populations. They argue that there is no support for age of onset as a decisive success factor in bilingual development, but they also stress the importance of age of onset as one of the factors interacting with the quantity and quality of the input. In recent work by Unsworth (Reference Unsworth, Belikova, Meroni and Umeda2007, Reference Unsworth2008), which is also on the acquisition of grammatical gender in Dutch, the role of the age of onset is crucially related to other factors involving the input, such as length of exposure.
In language development studies of (monolingual) children with SLI, factors such as quantity of input and length of exposure are generally not shown to play a role. In contrast, the factor age is an important factor, because the language growth of children with SLI is delayed in certain domains. For instance, Paradis shows that Rice (Reference Rice, Rice and Warren2004), who works within the context of a maturational model of language growth, argues that SLI children's problems with morphosyntactic phenomena such as finiteness marking lie mainly in timing (i.e., late onset). In this sense, then, as in bilingual development, age is also considered to be an important factor in studies of children with SLI. However, Paradis rightly stresses the importance of the observation that highly parallel profiles between L2 and SLI were found in a number of case studies, indicating that at a certain level of language development as measured by mean length of utterance, different populations of learners show highly similar profiles, even though they differ in other respects, including age.
The crucial point here is that even though the exact role played by age of onset in both the bilingual and (monolingual) SLI populations may differ, this factor is relevant in both cases; as Paradis convincingly shows, it cannot be interpreted independently of other factors. In this regard, it is interesting to note that in bilingual development the age of onset factor is closely related to other factors such as the (quantity and quality of the) input, and is often difficult to disentangle from them, but in the development of (monolingual) children with SLI the age of onset factor (of production rather than onset of exposure) crucially does not necessarily correlate with differences in input quantity but does correlate with other (more cognitive) factors.Footnote 1
SOCIETAL CONTEXTS
Paradis discusses work by Orgassa and Weerman (Reference Orgassa and Weerman2008) and Steenge (Reference Steenge2006) on Dutch bilingual SLI children. These authors have found extremely poor performance on Dutch grammatical gender morphology in these children, and explain this by assuming the children show a cumulative effect of bilingualism and SLI. Paradis does not agree with this explanation, and stresses the importance of social factors here. She writes that “it may take longer for bilinguals to catch up to their monolingual peers in contexts where bilingual children come from nonintegrated, socioeconomically disadvantaged minority groups.” We completely agree with her. In our own work with Cornips (Cornips & Hulk, Reference Cornips and Hulk2008), the acquisition of grammatical gender in Dutch by bilingual children from different sociolinguistic backgrounds is compared: (a) bilingual children from ethnic minority communities, a similar context to the SLI children studied by Orgassa and Weerman; (b) bilingual children from expatriate families, as studied by Unsworth (Reference Unsworth, Belikova, Meroni and Umeda2007, Reference Unsworth2008) and Hulk (Reference Hulk, Camacho, Flores-Ferrán, Sánchez, Déprez and Cabrera2007); and (c) bidialectal children (Cornips & Hulk, Reference Cornips, Hulk, Lefebvre C., White and Jourdan2006). Cornips and Hulk (Reference Cornips and Hulk2008) found an important difference in success rates between the bilingual children from the ethnic minority communities and the other bilingual children, which could not be related to factors such as age of onset. Moreover, they stressed the importance of the fact that the quality of the input to these children is different, because their parents are nontarget-like L2 speakers of Dutch themselves, and sociolinguistic research has shown that Dutch is often the home language in these families (cf. Jongenburger & Aarsen, Reference Jongenburger and Aarssen2001).
If we assume with Paradis that it is the societal context that is the main factor for failure in the bilingual SLI children studied by Orgassa and Weerman, we predict that bilingual SLI children from a more additive societal context (as Paradis calls it) will do much better on Dutch grammatical gender. We can think here of bilingual SLI children from English or French expatriate families living in The Netherlands, who generally have a very positive attitude toward bilingual education. Unfortunately, no research exists to date for this population, but it is certainly an interesting direction for future research.
Another societal aspect of studies as reviewed by Paradis is that they may contribute to the advice given to parents raising their children bilingually. Recently, a parliamentary committee in The Netherlands, which was faced with high percentages of failure at school by children from ethnic communities, blamed the bilingualism of these children as the main reason for their lack of success in the Dutch educational system. As one possible solution to this problem, the committee wondered whether it would not be better if the families in these communities were only to speak Dutch at home and not their other language. For these politicians, speaking two languages is considered too much of a burden for the children, an attitude that would be considered outdated by (most) linguists but which shows that, in The Netherlands at least, much work remains to be done in terms of the dissemination of research results regarding bilingualism in general and the bilingual child in particular. In this regard, it is very interesting to consider some of the Dutch sociolinguistic studies from the 1980s on bidialectal children.Footnote 2 For example, Kroon and Vallen (Reference Kroon and Vallen2004) found that children whose language at home was the dialect (i.e., the minority language) fared better at school on standard Dutch (i.e., the majority language) than the children whose families were told to switch to the majority language at home. Clearly, these results would lead to different advice than that suggested by the parliamentary committee. In relation to the question of what advice to give to immigrant families in a different, but comparable bilingual context in Canada, it is interesting that Paradis cites Genesee, Paradis, and Crago (Reference Genesee, Paradis and Crago2004), who found that switching to the majority language at home probably will not have a beneficial affect on the child's development of the majority language.
COGNITIVE AND LINGUISTIC FACTORS
We mentioned some of the similarities between L2 and SLI as reviewed by Paradis. Paradis also mentioned some differences that she found between these two profiles, relating these to neurocognitive maturity and dual language experience. For example, she shows that in studies on the acquisition of BE and inflectional tense morphemes in English, it was found that L2 children make more commissions errors and more BE overgeneralizations than children with SLI. Of interest, Paradis proposes that these errors signal a kind of creativity in L2 children's language use. She states that “it could be the case that when children encounter input in a target language when neurocognitively more mature, and with another language system already developing, this situation produces more creativity in their learning patterns.” Paradis views this creativity as something highly positive, and rightly so in our opinion.
Such creativity holds not only for children who start learning their language after age 3 or 4, but also for children who are raised bilingually from birth: it reminds us of what has been formulated in terms of “pooling of resources” and “‘short cuts” in 2L1 bilingual children by Gawlitzek and Tracy (Reference Gawlitzek-Maiwald and Tracy1996) in their seminal article on bilingual bootstrapping. Paradis also mentions bilingual bootstrapping as an example of “another potential compensatory mechanism in the organization of dual language systems in bilingual children.” She also suggests that L2 children's sensitivity to frequent and syntactically unmarked grammatical morphemes “might be heightened” or “in any case [be] stronger than the sensitivity of monolingual L1 learners.” Heightened sensitivity might be related to the more general cognitive aspect of enhanced metalinguistic awareness of bilingual children as found by Bialystok, among others. However, it could also be the result of the fact that bilingual children have access to and can activate two language systems, that they have a dual language organization, as Paradis calls it.
We found in our own work (Hulk & van der Linden, Reference Hulk and Van Der Linden2010) that Romance–Dutch bilingual children seemed to be highly sensitive to gender on definite determiners in Dutch, sometimes showing quicker development when compared to their monolingual peers. Paradis also stresses the positive effects of facilitative cross-linguistic interaction in the morphosyntactic development of bilingual children, with and without SLI. Just as in her other work, Paradis highlights the positive aspects of cross-linguistic effects here, contrary to what is often done in the literature. This again completely corresponds to our own view on early child bilingualism (Hulk & Mueller, Reference Hulk and Müller2000; Hulk & van der Linden, Reference Hulk and van der Linden1999). It is interesting that several other linguists working on early child bilingualism have also recently started to explicitly state such advantages (of bilingualism). For example, Tracy's (in press) recent article shows that “some young L2 learners may even outperform L1 learners in terms of speed” and Rothweiler (in press), citing Chilla (Reference Chilla2008), argues that “for SLI children learning a second language, their first language might be an advantage.”
METHODOLOGY
One of the other interesting aspects of Paradis' article is that she raises the question of how to interpret and compare the results of all these different bilingual populations and how to measure cumulative effects. Her suggestions again do not hold only for bilingual children with language disorders, but for bilingual populations in general. Paradis suggests that one way of measuring cumulative effects with sequential bilinguals is to examine the magnitude of difference between pairs of learner populations. For instance, she argues that if the magnitude effect, or effect size, is larger for TD bilinguals when compared with bilinguals with SLI than it is for TD monolinguals when compared with children with SLI, this could be seen as evidence for a cumulative effect in bilinguals. In applying this method to measuring the supposed effects in the data of the Dutch bilingual children with SLI from Orgassa and Weerman (see above, Reference Orgassa and Weerman2008), it is interesting that she did not find evidence for cumulative effects. We believe that the magnitude effect is an interesting methodology to use in future research in both SLI–TD comparisons or with children and comparisons of child and adult L2 learners.
Paradis also found that the amount of exposure to the L2 matters for comparisons between different bilingual groups to be interpretable. This again stresses the importance of taking into account as many factors as possible and using a suitable methodology for interpreting and comparing the results of different bilingual populations before drawing any definite conclusions about atypicalities in language development, both theoretically and clinically.
CONCLUSION
In this Commentary on Paradis' Keynote Article, we have taken her question of whether bilingualism can be viewed as a kind of “therapy” as a starting point and looked for elements to interpret and compare linguistic data from different early child bilingual populations and their monolingual peers. We particularly welcome the positive, and careful and methodologically strict approach Paradis takes when reviewing research on bilingual children. We share her vision that having an existing linguistic system alters the representational knowledge and procedural mechanisms used to acquire another. We also completely agree that the societal context plays a more important role than has sometimes been claimed and has to be taken into account.
With respect to the more practical aspects of bilingualism research and its implications, one of the important lessons to learn from this article is that Paradis shows in a very clear way that there is no “one size fits all” answer to give to the frequently asked question of how to successfully raise bilingual children. The path to successful bilingual acquisition is long and winding, and littered with complicating factors, factors that differ both within and across groups. Paradis not only highlights the complexity of this path but also makes useful and directly implementable suggestions on how to try and disentangle the relevant factors. In particular, we hope that her suggestion about an alternative way of measuring cumulative effects will be taken up in future research with bilingual children, both with and without SLI.
You have
Access
In her very interesting Keynote Article, Johanne Paradis gives a clear overview of recent research at the interface of bilingual development and child language disorders, and highlights its theoretical and clinical implications. She raises the challenging question of “whether bilingualism can be viewed as a kind of ‘therapy’ for SLI.” At first sight, this is perhaps a surprising question, because one of the predominant views in the literature is that bilingual children with specific language impairment (SLI) will exhibit difficulties and perhaps a “double delay.” It is this challenging question that we consider in more detail here.
More specifically, we would like to take a closer look at what Paradis' perspective tells us about the opportunities and vulnerabilities in early child bilingualism in general, for if bilingualism can be advantageous for children with SLI, it follows that it can also be advantageous for typically developing (TD) bilingual children. Paradis, who is one of the few linguists performing research on the interface of bilingual development and language disorders, systematically stresses the positive aspects and the advantages of bilingualism in different child populations. She also rightly reminds us that children who learn two languages before puberty are the majority worldwide.
Paradis furthermore stresses the importance of comparing different bilingual populations to learn more about the possible interaction between factors of success and failure in bilingual development. On the basis of her discussion, we can distinguish four domains in which bilingual child populations may generally differ: (a) age of onset, (b) societal contexts, (c) cognitive maturity, and (d) languages involved.
In the following we will attempt to disentangle the different factors of success and failure mentioned by Paradis in relation to the development of bilingual children with SLI, and try to establish what they tell us more generally about early child bilingualism in different populations.
AGE OF ONSET
In child bilingualism studies, differences in exposure patterns are related to well-known factors such as the age of onset of acquisition, input quantity, length of exposure, and language dominance. In these studies (e.g., Meisel, Reference Meisel2007; Montrul, Reference Montrul2008; Paradis, Reference Paradis, Hoff and Shatz2007), a distinction is often made between children growing up bilingually from birth, who are simultaneously bilingual (two first languages [2L1]) children, and sequential bilingual children who start acquiring their second language later (cL2), say, after age 3 (e.g., MacLaughlin, Reference McLaughlin1978). Paradis convincingly argues that such a distinction based on the age of onset can never be the only factor to take into account when explaining differences in bilingual language development. Cornips and Hulk (Reference Cornips and Hulk2008) make a similar point in a recent article on the acquisition of grammatical gender in Dutch in different bilingual populations. They argue that there is no support for age of onset as a decisive success factor in bilingual development, but they also stress the importance of age of onset as one of the factors interacting with the quantity and quality of the input. In recent work by Unsworth (Reference Unsworth, Belikova, Meroni and Umeda2007, Reference Unsworth2008), which is also on the acquisition of grammatical gender in Dutch, the role of the age of onset is crucially related to other factors involving the input, such as length of exposure.
In language development studies of (monolingual) children with SLI, factors such as quantity of input and length of exposure are generally not shown to play a role. In contrast, the factor age is an important factor, because the language growth of children with SLI is delayed in certain domains. For instance, Paradis shows that Rice (Reference Rice, Rice and Warren2004), who works within the context of a maturational model of language growth, argues that SLI children's problems with morphosyntactic phenomena such as finiteness marking lie mainly in timing (i.e., late onset). In this sense, then, as in bilingual development, age is also considered to be an important factor in studies of children with SLI. However, Paradis rightly stresses the importance of the observation that highly parallel profiles between L2 and SLI were found in a number of case studies, indicating that at a certain level of language development as measured by mean length of utterance, different populations of learners show highly similar profiles, even though they differ in other respects, including age.
The crucial point here is that even though the exact role played by age of onset in both the bilingual and (monolingual) SLI populations may differ, this factor is relevant in both cases; as Paradis convincingly shows, it cannot be interpreted independently of other factors. In this regard, it is interesting to note that in bilingual development the age of onset factor is closely related to other factors such as the (quantity and quality of the) input, and is often difficult to disentangle from them, but in the development of (monolingual) children with SLI the age of onset factor (of production rather than onset of exposure) crucially does not necessarily correlate with differences in input quantity but does correlate with other (more cognitive) factors.Footnote 1
SOCIETAL CONTEXTS
Paradis discusses work by Orgassa and Weerman (Reference Orgassa and Weerman2008) and Steenge (Reference Steenge2006) on Dutch bilingual SLI children. These authors have found extremely poor performance on Dutch grammatical gender morphology in these children, and explain this by assuming the children show a cumulative effect of bilingualism and SLI. Paradis does not agree with this explanation, and stresses the importance of social factors here. She writes that “it may take longer for bilinguals to catch up to their monolingual peers in contexts where bilingual children come from nonintegrated, socioeconomically disadvantaged minority groups.” We completely agree with her. In our own work with Cornips (Cornips & Hulk, Reference Cornips and Hulk2008), the acquisition of grammatical gender in Dutch by bilingual children from different sociolinguistic backgrounds is compared: (a) bilingual children from ethnic minority communities, a similar context to the SLI children studied by Orgassa and Weerman; (b) bilingual children from expatriate families, as studied by Unsworth (Reference Unsworth, Belikova, Meroni and Umeda2007, Reference Unsworth2008) and Hulk (Reference Hulk, Camacho, Flores-Ferrán, Sánchez, Déprez and Cabrera2007); and (c) bidialectal children (Cornips & Hulk, Reference Cornips, Hulk, Lefebvre C., White and Jourdan2006). Cornips and Hulk (Reference Cornips and Hulk2008) found an important difference in success rates between the bilingual children from the ethnic minority communities and the other bilingual children, which could not be related to factors such as age of onset. Moreover, they stressed the importance of the fact that the quality of the input to these children is different, because their parents are nontarget-like L2 speakers of Dutch themselves, and sociolinguistic research has shown that Dutch is often the home language in these families (cf. Jongenburger & Aarsen, Reference Jongenburger and Aarssen2001).
If we assume with Paradis that it is the societal context that is the main factor for failure in the bilingual SLI children studied by Orgassa and Weerman, we predict that bilingual SLI children from a more additive societal context (as Paradis calls it) will do much better on Dutch grammatical gender. We can think here of bilingual SLI children from English or French expatriate families living in The Netherlands, who generally have a very positive attitude toward bilingual education. Unfortunately, no research exists to date for this population, but it is certainly an interesting direction for future research.
Another societal aspect of studies as reviewed by Paradis is that they may contribute to the advice given to parents raising their children bilingually. Recently, a parliamentary committee in The Netherlands, which was faced with high percentages of failure at school by children from ethnic communities, blamed the bilingualism of these children as the main reason for their lack of success in the Dutch educational system. As one possible solution to this problem, the committee wondered whether it would not be better if the families in these communities were only to speak Dutch at home and not their other language. For these politicians, speaking two languages is considered too much of a burden for the children, an attitude that would be considered outdated by (most) linguists but which shows that, in The Netherlands at least, much work remains to be done in terms of the dissemination of research results regarding bilingualism in general and the bilingual child in particular. In this regard, it is very interesting to consider some of the Dutch sociolinguistic studies from the 1980s on bidialectal children.Footnote 2 For example, Kroon and Vallen (Reference Kroon and Vallen2004) found that children whose language at home was the dialect (i.e., the minority language) fared better at school on standard Dutch (i.e., the majority language) than the children whose families were told to switch to the majority language at home. Clearly, these results would lead to different advice than that suggested by the parliamentary committee. In relation to the question of what advice to give to immigrant families in a different, but comparable bilingual context in Canada, it is interesting that Paradis cites Genesee, Paradis, and Crago (Reference Genesee, Paradis and Crago2004), who found that switching to the majority language at home probably will not have a beneficial affect on the child's development of the majority language.
COGNITIVE AND LINGUISTIC FACTORS
We mentioned some of the similarities between L2 and SLI as reviewed by Paradis. Paradis also mentioned some differences that she found between these two profiles, relating these to neurocognitive maturity and dual language experience. For example, she shows that in studies on the acquisition of BE and inflectional tense morphemes in English, it was found that L2 children make more commissions errors and more BE overgeneralizations than children with SLI. Of interest, Paradis proposes that these errors signal a kind of creativity in L2 children's language use. She states that “it could be the case that when children encounter input in a target language when neurocognitively more mature, and with another language system already developing, this situation produces more creativity in their learning patterns.” Paradis views this creativity as something highly positive, and rightly so in our opinion.
Such creativity holds not only for children who start learning their language after age 3 or 4, but also for children who are raised bilingually from birth: it reminds us of what has been formulated in terms of “pooling of resources” and “‘short cuts” in 2L1 bilingual children by Gawlitzek and Tracy (Reference Gawlitzek-Maiwald and Tracy1996) in their seminal article on bilingual bootstrapping. Paradis also mentions bilingual bootstrapping as an example of “another potential compensatory mechanism in the organization of dual language systems in bilingual children.” She also suggests that L2 children's sensitivity to frequent and syntactically unmarked grammatical morphemes “might be heightened” or “in any case [be] stronger than the sensitivity of monolingual L1 learners.” Heightened sensitivity might be related to the more general cognitive aspect of enhanced metalinguistic awareness of bilingual children as found by Bialystok, among others. However, it could also be the result of the fact that bilingual children have access to and can activate two language systems, that they have a dual language organization, as Paradis calls it.
We found in our own work (Hulk & van der Linden, Reference Hulk and Van Der Linden2010) that Romance–Dutch bilingual children seemed to be highly sensitive to gender on definite determiners in Dutch, sometimes showing quicker development when compared to their monolingual peers. Paradis also stresses the positive effects of facilitative cross-linguistic interaction in the morphosyntactic development of bilingual children, with and without SLI. Just as in her other work, Paradis highlights the positive aspects of cross-linguistic effects here, contrary to what is often done in the literature. This again completely corresponds to our own view on early child bilingualism (Hulk & Mueller, Reference Hulk and Müller2000; Hulk & van der Linden, Reference Hulk and van der Linden1999). It is interesting that several other linguists working on early child bilingualism have also recently started to explicitly state such advantages (of bilingualism). For example, Tracy's (in press) recent article shows that “some young L2 learners may even outperform L1 learners in terms of speed” and Rothweiler (in press), citing Chilla (Reference Chilla2008), argues that “for SLI children learning a second language, their first language might be an advantage.”
METHODOLOGY
One of the other interesting aspects of Paradis' article is that she raises the question of how to interpret and compare the results of all these different bilingual populations and how to measure cumulative effects. Her suggestions again do not hold only for bilingual children with language disorders, but for bilingual populations in general. Paradis suggests that one way of measuring cumulative effects with sequential bilinguals is to examine the magnitude of difference between pairs of learner populations. For instance, she argues that if the magnitude effect, or effect size, is larger for TD bilinguals when compared with bilinguals with SLI than it is for TD monolinguals when compared with children with SLI, this could be seen as evidence for a cumulative effect in bilinguals. In applying this method to measuring the supposed effects in the data of the Dutch bilingual children with SLI from Orgassa and Weerman (see above, Reference Orgassa and Weerman2008), it is interesting that she did not find evidence for cumulative effects. We believe that the magnitude effect is an interesting methodology to use in future research in both SLI–TD comparisons or with children and comparisons of child and adult L2 learners.
Paradis also found that the amount of exposure to the L2 matters for comparisons between different bilingual groups to be interpretable. This again stresses the importance of taking into account as many factors as possible and using a suitable methodology for interpreting and comparing the results of different bilingual populations before drawing any definite conclusions about atypicalities in language development, both theoretically and clinically.
CONCLUSION
In this Commentary on Paradis' Keynote Article, we have taken her question of whether bilingualism can be viewed as a kind of “therapy” as a starting point and looked for elements to interpret and compare linguistic data from different early child bilingual populations and their monolingual peers. We particularly welcome the positive, and careful and methodologically strict approach Paradis takes when reviewing research on bilingual children. We share her vision that having an existing linguistic system alters the representational knowledge and procedural mechanisms used to acquire another. We also completely agree that the societal context plays a more important role than has sometimes been claimed and has to be taken into account.
With respect to the more practical aspects of bilingualism research and its implications, one of the important lessons to learn from this article is that Paradis shows in a very clear way that there is no “one size fits all” answer to give to the frequently asked question of how to successfully raise bilingual children. The path to successful bilingual acquisition is long and winding, and littered with complicating factors, factors that differ both within and across groups. Paradis not only highlights the complexity of this path but also makes useful and directly implementable suggestions on how to try and disentangle the relevant factors. In particular, we hope that her suggestion about an alternative way of measuring cumulative effects will be taken up in future research with bilingual children, both with and without SLI.