2.1.1 Executive Function (Control)

Bialystok, Craik and Luk (Reference Bialystok, Craik and Luk2012: 241) describe executive control as “the set of cognitive skills based on limited cognitive resources for such functions as inhibition, switching attention, and working memory.” Bilingualism has been hypothesized to impact executive control positively, since bilinguals, when using one language, unconsciously need to inhibit the other. This follows from the assumption that the languages in the bilingual’s mind are typically coactive (see Aronin & Singleton Reference Aronin and Singleton2012: 85 for discussion). Accordingly, executive control experiences a training effect. The following quotation offers an apt summary of this connection (see also Bialystok Reference Bialystok, Miller, Bayram, Rothman and Serratrice2018: 294):

Executive function is assessed using a variety of tasks, notably the Simon task, the flanker task, and the Stroop task (Hilchey & Klein Reference Hilchey and Klein2011: 628). These comprise experiments involving congruent and incongruent conditions of certain stimuli (colours, shapes, letters, etc.) while gauging the performance (usually reaction times) of subjects when confronted with these conditions (Donnelly, Brooks & Homer Reference Donnelly, Brooks, Homer, Noelle, Dale, Warlaumont, Yoshimi, Matlock, Jennings and Maglio2015: 596). For example, when doing the Simon task, participants see differently coloured (red or green) shapes (circles, squares) on either the left or right half of a computer screen. The colours are associated with buttons equally positioned on the left- or right-hand side. Participants are then asked to press the button associated with the relevant colour as soon as the object appears on the screen. If the coloured object and its associated button match in position (both left or right in the congruent condition), the measured reaction time is typically lower than if they do not (one left and the other right in the incongruent condition). Inhibitory control is defined in terms of reaction time in the incongruent condition minus reaction time measured in the congruent condition (interference cost). Smaller differences are interpreted as greater inhibitory control. Global reaction time is calculated as the mean of both congruent and incongruent reaction times (Donnelly, Brooks & Homer Reference Donnelly, Brooks, Homer, Noelle, Dale, Warlaumont, Yoshimi, Matlock, Jennings and Maglio2015: 596; Paap, Johnson & Sawi Reference Paap, Johnson and Sawi2015: 267). The flanker task and the Stroop task are variations on this experimental set-up. Arrows pointing in one direction are flanked by an arrow pointing in the same (congruent) or the opposite (incongruent) direction. The Stroop task is based on colour adjectives printed in matching and non-matching colours.

It may be worth pointing out that these are standard psychological tests that have been used to examine the influences of a host of other factors on executive control, such as level of blood alcohol, tiredness, illnesses, and so on. In studies searching for multilingual advantages, a bilingual subject group is typically tested against a monolingual control group. The tests described earlier have been conducted with both children and adults. They tend to report faster (i.e. ‘better’) performance in bilinguals over monolinguals. There are interesting age differences such that, for example, young adults show the smallest differences across monolingual and bilingual groups or no differences at all (Bialystok Reference Bialystok, Miller, Bayram, Rothman and Serratrice2018: 292–293).

2.1.2 Cognitive Reserve

More recently, it has also been hypothesized that bilingualism positively impacts cognitive reserve. This construct describes a cognitive buffer that shields against age-related symptoms of cognitive decline. It is widely accepted that physical, intellectual, and social engagement contribute to cognitive reserve (Bialystok, Craik & Luk Reference Bialystok, Craik and Luk2012: 246–247). The impact of bilingualism on cognitive reserve was, inter alia, tested on a sample of 184 patients diagnosed with dementia (91 monolinguals and 93 bilinguals) by Bialystok, Craik and Freedman (Reference Bialystok, Craik and Freedman2007). The study found that patients with a bilingual background showed a delayed onset in the symptoms of dementia, appearing approximately four years later than in monolingual patients. The analysis controlled for several confounding factors such as education and occupation status. Similarly, Craik, Bialystok and Freedman (Reference Craik, Bialystok and Freedman2010) examined the effects of bilingualism on the onset of Alzheimer’s disease (102 bilinguals and 109 monolinguals). They also reported a significant delay in the onset of the relevant symptoms in bilingual patients. Aveledo et al. (Reference Aveledo, Higueras, Marinis, Bose, Pliatsikas, Meldaña-Rivera, Martínez-Ginés, García-Domínguez, Lozano-Ros, Cuello and Goicochea-Briceño2021) offer an extension of this line of research into multiple sclerosis. In relation to age differences, Bialystok et al. (Reference Bialystok, Craik, Klein and Viswanathan2004) encountered stronger processing advantages in older bilinguals relative to older monolinguals when comparing them to corresponding younger cohorts on the Simon task. In sum, it is argued that bilingualism can counterbalance age-related cognitive decline.

2.1.3 Controversial Issues

After substantial initial enthusiasm regarding bilingual boosting effects on executive control, usually measured in terms of higher inhibitory control and shorter general reaction times, the field has been going through a sobering period that started around 2010. As positive effects associated with bilingualism failed to replicate in follow-up studies, recent publications have called for a reorientation of the field (Leivada et al. Reference Leivada, Westergaard, Duñabeitia and Rothman2020; Paap & Greenberg Reference Paap and Greenberg2013; Paap, Johnson & Sawi Reference Paap, Johnson and Sawi2016). The following paragraphs summarize the main points of criticism.

As it is impossible even to attempt to survey and discuss the plethora of studies on bilingualism and executive control, I here exclusively rely on the available meta-analyses. Even these need to be interpreted with caution, as they can only compare subsamples from the pool of potentially relevant studies, use diverging criteria for inclusion and exclusion, and focus on different phenomena.Footnote ⁴ For instance, Olusola et al. (Reference Olusola, Lavin, Thompson and Ungerleider2010: 211–213) considered no fewer than 5,185 articles for inclusion in their meta-analysis, but eventually confined the comparison to 63 studies from 39 articles involving 6,022 participants, since only these met their selection criteria.Footnote ⁵ Hilchey and Klein (Reference Hilchey and Klein2011) used 31 studies culled from 13 publications. Donnelly, Brooks and Homer (Reference Donnelly, Brooks, Homer, Noelle, Dale, Warlaumont, Yoshimi, Matlock, Jennings and Maglio2015) reported from 39 studies with 73 comparisons and 5,538 participants. Paap, Johnson and Sawi (Reference Paap, Johnson and Sawi2015) restricted their meta-analysis to studies conducted after 2011. De Bruin, Treccani and Della Sala (Reference de Bruin, Treccani and Sala2015) pursued the publication trajectory of 128 studies from conference presentation to final publication in print. Lehtonen et al. (Reference Lehtonen, Soveri, Laine, Järvenpää, de Bruin and Antfolk2018) compared 152 studies, including unpublished theses.

Perhaps the most important points of concern are sample size and recruitment processes (see also Bialystok Reference Bialystok2020: 13). It is not at all uncommon to find samples below 100 or even 50 participants. Given that these are meant to represent the totality of bilingual and monolingual speakers, some scepticism appears warranted. Moreover, it is well known that small samples can produce strong statistical cohort effects. Leivada et al. (Reference Leivada, Westergaard, Duñabeitia and Rothman2020: 8) report a recent small increase in average sample size. Paap, Johnson and Sawi (Reference Paap, Johnson and Sawi2015: 267, 271) argue that studies claiming bilingual advantages typically rely on small samples, whereas bigger samples tend to reveal no effects. The most common participants for studies are students. This, of course, is due to various practical considerations, but students are certainly not representative of society at large.

There may also be a problem of publication bias, as small studies reporting strong positive effects may stand higher chances of being accepted in journals. De Bruin, Treccani and Della Sala (Reference de Bruin, Treccani and Sala2015: 99) make exactly this point. According to their survey of eventually published conference abstracts, “[s]tudies with results fully supporting the bilingual-advantage theory were most likely to be published, followed by studies with mixed results. Studies challenging the bilingual advantage were published the least.” Lehtonen et al. (Reference Lehtonen, Soveri, Laine, Järvenpää, de Bruin and Antfolk2018: 413) come to similar conclusions.Footnote ⁶ Paap, Johnson and Sawi (Reference Paap, Johnson and Sawi2015: 266) point out that null findings may easily be interpreted as type II errors by reviewers, leading to a rejection of the studies.Footnote ⁷

A third problem area relates to various confounding background variables, the most important of which is the distinction between monolinguals and bilinguals. Traditionally conceived as a categorical binary variable, more recent research has revealed that bilingualism is a continuum or rather spectrum of experiences (DeLuca et al. 2019; Leivada et al. Reference Leivada, Westergaard, Duñabeitia and Rothman2020: 6–7).Footnote ⁸ Even the otherwise helpful distinction between balanced and unbalanced bilinguals represents more of a label for a problem instead of a solution. Speakers can be bilingual in many different ways and further studies would need to operationalize and assess this correctly and consistently. We will encounter many ways of being bilingual and multilingual in the later chapters of this book. Additional confounds include the languages involved and their social prestige, age, socio-economic status, cognitive development, research lab, task design, code-switching behaviour, and even cultural differences (Paap, Johnson & Sawi Reference Paap, Johnson and Sawi2015: 268).Footnote ⁹ One may add that bilinguals may also maintain different individual conceptions of their bilingualism (double monolingualism versus one language system) that may tinker with the measurable effects. Finally, there remains the problem of causality (does bilingualism impact cognition or vice versa?) and the thorny issue of using appropriate statistics.

2.1.4 Summary

The scientific quest for bilingual disadvantages and advantages in relation to cognition has now been going on for more than a hundred years. Considering the entire time span, it is fair to say that it has produced mixed results with ups and downs, winners and losers (see also Section 1.4.1). In view of the research effort (manpower, lab time, etc.) spent, these inconclusive results are surprising. The quest is likely to continue for another hundred years, but my suspicion at this point is that there will be no conclusive outcome. It may therefore be wise to pause for a moment and ask what drives this endeavour and what additional complications there are. I can think of at least the following three points.

First, the issue of bilingual advantages and multilingualism in general is intimately tied to the allocation of resources. I am not aware of any exact calculations, but I presume that it is more costly to organize a society on bilingual or multilingual principles in comparison to a simple monolingual form of organization. Central systems of society, such as education, administration, medical services, and so on, need to be able to respond to different language demands. A bitter technocrat may argue that this is simply too expensive. In the long run, however, it may turn out that such resources are well spent, as the devaluation and loss of linguistic identities represent social time bombs that can heftily backfire, taking societies to the brink of collapse (see Section 1.4.3). The Canadian model of integration stands out as one of the more far-sighted in this respect.

Second, scholars and their students may – consciously or unconsciously – participate in this ideological debate, which, in turn, may influence how studies are designed and hence their outcomes. For example, I have frequently noticed how sensitively students adjust to my points of view. Scholars who believe in multilingualism, bilingual advantages, and the beauty of linguistic diversity are likely to influence their students in this way, or they unconsciously attract students who harmonize with their views. Since study participants tend to be recruited from students, some ideological bias appears unavoidable. Bilingual students may be highly motivated in such an environment, whereas monolingual students may merely participate for the sake of the incentives.

Third, the very idea of what counts as an advantage is subject to various, partially conflicting interpretations. For example, although the aforementioned increase in cognitive reserve associated with bilingualism appears to be a marvel, Leivada et al. (Reference Leivada, Westergaard, Duñabeitia and Rothman2020: 4) argue that a delayed appearance of phenomena of neurodegeneration may be tantamount to a delayed diagnosis, as the patients simply compensate behaviourally.

2.2.1 Language, Competence, and Performance

These terminological issues are quite critical, as the terms ‘language’ and ‘competence’ can be understood in fundamentally different ways. When linguists – especially in acquisition studies and bilingualism research – talk about language, they typically refer to the underlying mental capacity or what Chomsky (Reference Chomsky1986) defined as the “I-language” (in opposition to “E-language”). Except for certain clinical cases, all human beings are believed to possess the mental capacities necessary to develop languages (a language capacity).Footnote ¹³ These initial mental representations (“initial state”) then further mature according to the linguistic input children receive from their environment. Input from different languages provokes different maturation processes, but the linguistic core is universal and remains the same. Crucially, all children – provided they receive the requisite input – develop into fully competent speakers (native speakers). There are no better or worse speakers.

The perspective of education studies, with their focus on academic language, is completely different, as academic language is a highly specific register systematically developed throughout our academic careers (Biber Reference Biber2006). Some parents start this process even before the beginning of school. It appears to be a matter of debate how early, in which contexts, and in which quality and quantity even very young children acquiring their first language receive instructions related to academic language from their caretakers – and how they interpret them. Beginning at school, at the latest, children develop literacy and are explicitly taught the rules of written languages. They learn the differences between spoken and written as well as appropriate and inappropriate language. They begin to develop explicit grammatical knowledge and metalinguistic awareness, and they are penalized for using the wrong code in the wrong context. Moreover, the academic world encourages spoken language according to the principles of written languages (“Speak in full sentences!”).Footnote ¹⁴ There is substantial prescriptivism, of which children and also many educators are unaware, working in the background. Opposing language “fluent surface skills” to “complex cognitive operations” (Cummins Reference Cummins1979: 227), the educational perspective implies a devaluation of basic interpersonal communicative skills and a concomitant valorization of cognitive academic language proficiency.Footnote ¹⁵ In this world, people are sorted into more and less educated according to their language use. Part of the mission of the education system is to produce people with different levels of education. Here, we do find better and worse speakers or language users.

Similar observations are relevant for the term ‘competence’, as linguists use this term for the knowledge system behind languages that cannot be observed directly. In education studies, the same term is widely used to gauge performance, that is, the observable output of language processing. Whereas (mature) native speakers of a language are, by definition, equally competent from a linguistic perspective, the educational approach imposes a finely differentiated grid of competence levels on the language users. Our naturally acquired language competence is different from the culturally developed language competence – especially that found in the education system. Needless to say, these differences in perspective offer endless opportunities for misunderstandings and fruitless controversial discussions.

In several successive publications, Cummins (Reference Cummins1976, Reference Cummins1979, amongst others) developed two hypotheses on academic language competence in the bilingual speaker, namely the “threshold hypothesis” and the “interdependence hypothesis” (see also Baker Reference Baker2011). They both concern the level of academic language competence of the languages in the bilingual mind. They will be briefly discussed in what follows.

2.2.2 Language Competence Levels (Thresholds)

Bilingual education programmes revealed that minority language children tend to underperform regarding educational attainment and academic language competence, whereas majority language children appear to benefit from bilingualism (Cummins Reference Cummins1979). Minority language children are members of indigenous communities, prominently in the United States, Canada, and Australia, who were raised in their community language and then entered English-medium educational programmes. Alternatively, the label is used for immigrant children – no matter whether foreign-born or not – who equally start educational programmes conducted in the majority language English. Immigrant children in other contexts (such as Europe) face similar problems in other majority language situations (German, French, Swedish, etc.). In contrast, bilingual majority language children acquire another language in addition to the language of their environment, as in the context of English–French bilingualism in Canada, but also majority language children in Europe learning English at an early age.

Various additional conditions correlate with this distinction. Minority language children typically go through transitional bilingual submersion programmes or simply majority language submersion programmes. They are considered to develop subtractive bilingualism or even “semilingualism” (see below). Subtractive bilingualism can lead to dominant bilingualism where high academic competence is only achieved in one language, while the other language remains at a lower competence level. Bilingual majority language children attend immersion programmes, enjoy maintenance bilingualism, and become additive bilinguals. The outcome of additive bilingualism is high levels of (academic) competence in both languages (balanced bilingualism). Moreover, the two groups tend to differ strongly in terms of social background and socio-economic status. Finally, the languages involved carry very different levels of social prestige, with minority languages, typically, being devalued in the respective majority language contexts. Minority languages may be oral languages without a writing system.

The threshold hypothesis (Cummins Reference Cummins1979: 230) claims that situations of additive bilingualism produce positive cognitive effects, whereas such positive effects cannot be expected in situations of dominant bilingualism (in which one language is stronger than the other). In other words, a certain level of academic language competence needs to be surpassed in both languages before any “positive cognitive effects” of bilingualism can be detected. As Cummins explains,

Cummins thus postulates a relationship between linguistic competence in first and second language and cognitive growth. The definition of “positive cognitive effects” remains somewhat unclear, but they can be expected to relate to higher educational attainment or better school grades (see also Lasagabaster Reference Lasagabaster1998, Reference Lasagabaster2001). Put differently, additive bilingualism is associated with higher intelligence. As this is a far-reaching proposition, some scepticism appears to be in order. As for bilingual children with low academic language competence in both languages, Cummins (Reference Cummins1979: 230) even assumes negative cognitive effects or intellectual disability (retardation). This is the widely discussed condition of “semilingualism” (Swedish halvspråkighet; Hansegard Reference Hansegard1968). It justly attracted considerable criticism due to its pejorative implications. Although semilingualism as such defines a condition of low academic language competence in the two languages of the bilingual child, there is the danger of mistaking it for language ability in general, thus creating linguistically deficient children. As MacSwan warns us,

The conditions outlined above produce a lower and a higher threshold of bilingual competence that can vary from child to child and cannot be defined in absolute terms (Cummins Reference Cummins1979: 230). This makes it difficult to operationalize them and produce empirical support for them. Berthele and Vanhove (Reference Berthele and Vanhove2020: 553) point out that “no one so far has shown empirically that there is a non-linear developmental slope whose shape would be evidence for such thresholds.”

2.2.3 Language Interdependence

Furthermore, Cummins (1979) hypothesized a relationship of interdependence between the languages in the bilingual mind such that academic competence in the one language positively influences academic competence in the other. This interdependence hypothesis, provided it is valid, has important practical implications. For example, heritage language speaking children could benefit from developing their heritage language academically inasmuch as this could be advantageous for academic development in the majority language. As Cummins explains,

The interdependence of languages in the sense of Cummins (Reference Cummins1979; see also 1976) can be viewed as a special case of facilitative cross-linguistic influence in which proficiency in one language is assumed to boost proficiency in another (Berthele & Vanhove Reference Berthele and Vanhove2020: 550–551; Lechner & Siemund Reference Lechner and Siemund2014). It concerns largely unspecified higher-order cognitive abilities to the exclusion of the more specific linguistic properties that are the traditional focus of studies charting cross-linguistic influence (see Chapter 3).Footnote ¹⁶ Although interdependence was originally conceptualized in the context of bilingualism proper, it was more recently extended to multilingual situations, especially the acquisition of English as a third language on a previous bilingual experience (see Chapter 4).

As with thresholds, it remains challenging to operationalize and measure the interdependence between languages. As Berthele and Vanhove (Reference Berthele and Vanhove2020: 553–554) point out, several in principle observable data sets measuring academic competence in heritage and majority language are compatible with language interdependence, especially under the assumption of thresholds. What is more, language interdependence is also consistent with several alternative principles such as language giftedness or more general cognitive skills (working memory, general intelligence, world knowledge, test-wiseness; Berthele & Vanhove Reference Berthele and Vanhove2020: 562).

2.2.4 Summary

Language acquisition and language development are two different pairs of shoes, although the latter term may encompass the former (as in this book). To use an analogy, their relationship is comparable to that of walking and dancing. Children learn to walk without instruction, and this is what also happens with language during first language acquisition. Dancing is a social activity that requires instruction and training – similar to the development of academic language in the education system. Evidently, walking is a prerequisite for dancing and the more fundamental psycho-motoric activity. It appears plausible that previous dancing experience facilitates the mastering of novel dances, much like knowledge of the academic conventions in one language facilitates the learning of the relevant conventions in another. Talent for dancing corresponds to verbal intelligence. Viewed from this perspective, language interdependence is not at all unexpected.

Moreover, learning the conventions of academic language is not the same as learning an additional language, as they relate to different, albeit partially overlapping knowledge systems. This explains (part of) the struggle experienced by minority language children during their early school careers. They are faced with a double challenge, while majority language children can focus on the academic content. Transitional bilingual programmes provide children with the necessary time to learn the majority language:

However, neither language interdependence nor bilingual programmes will be able to counterbalance low language prestige and the concomitant loss of motivation to invest into minority languages. The reasons behind these problems are entirely social. They do not arise with prestigious languages. Across the world, English is learnt as an additional language with a continuously decreasing age of onset. The result is widespread bilingualism of the relevant national language and English. In the Scandinavian countries, the Netherlands, and increasingly also Germany, it is becoming more and more difficult to find monolingual speakers. To the best of my knowledge, no one has ever associated this situation of bilingualism with cognitive effects, neither in terms of retardation or advancement. Learning English is simply a social requirement, there being no room to assess its potentially negative or positive consequences. In addition, it would be practically impossible to recruit the necessary monolingual controls to drive any such comparison.

On a similar note, bilingualism across Europe – involving a vernacular in combination with the highly prestigious lingua francas of Latin, French, and now English – has always been considered an emblem of high education largely restricted to the elites. If at all, there is a narrative of high education corresponding to cognitive advancement in an educational sense. No one ever suspected this condition to produce mental retardation. As education is a hallmark of contemporary knowledge-based societies, this type of bilingualism (national language plus English) is now disseminating through societies all over the world.

2.3.1 Types of Metalinguistic Awareness

Gombert (Reference Gombert1992) introduced the very useful opposition of epilinguistic and metalinguistic awareness, with epilinguistic awareness referring to unconscious, spontaneous, and contextualized manifestations of language reflection as self-repair in speech (see also Wrembel Reference Wrembel, Roehr and Ganem-Gutierrez2013: 120). Self-repair of slips of the tongue would fall into this domain. Metalinguistic awareness proper, in contrast, is decontextualized, conscious, and intentional, a typical example being language analysis in the classroom or conscious reflection on language properties. Most work on metalinguistic awareness concerns this conscious type.

Metalinguistic awareness as such may concern all levels of linguistic analysis. It can be developed and tested in relation to graphemes, phonemes, morphemes, words, phrases, clauses, sentences, registers, and text types – essentially any linguistic unit that is amenable to conscious inspection (see Sanz Reference Sanz and Chapelle2012 for an overview). To be sure, professional linguists develop further types of metalinguistic awareness, but this is subject to more extensive theorizing and also speech technology. Metalinguistic awareness typically reveals itself by subjects’ comments on and analyses of the above language domains. Such comments and analyses may be induced by metalinguistic awareness tests (such as Pinto, Titone & Trusso Reference Pinto, Titone and Trusso1999). During second and further language acquisition, metalinguistic awareness may follow the recognition of errors, divergent hypotheses about the target language, as well as conscious comparisons with earlier acquired languages. One may further distinguish metapragmatic awareness (Safont Jordà Reference Safont Jordà2003, Reference Safont Jordà2005) that reveals conventions about language use unconsciously adhered to (as, for example, concerning the use of the imperative; see Siemund Reference Siemund2018b).

2.3.2 Metalinguistic Awareness and Language Proficiency

Multilingualism, metalinguistic awareness, and language proficiency (in an academic sense) are highly interrelated. This would not appear to be entirely unexpected. The acquisition and development of languages in the education system proceeds on the basis of reflective pedagogies that themselves are based on complex linguistic concepts and theorizing – although this theorizing and the concepts are hardly ever questioned in the school system. One may suspect that students performing well on language proficiency tasks (oral and written production and reception) also acquired the underlying linguistic concepts and are able to reflect on them (and vice versa). In multilingual situations – especially those embedded in the education system – substantial time is devoted to teaching languages including the requisite linguistic concepts. Here, too, it would appear reasonable to surmise a positive relationship between (educational) multilingualism, metalinguistic awareness, and language proficiency.

Such positive correlations have been documented in several studies. For example, in a comprehensive study carried out in the Basque Country with 252 bilingual students (Basque, Spanish) who studied English as a foreign language, Lasagabaster (Reference Lasagabaster1998, Reference Lasagabaster2001) identifies significant correlations between metalinguistic awareness, language proficiency, and various skills in English (listening, reading, speaking, grammar, and writing). In particular, Lasagabaster (Reference Lasagabaster1998) demonstrates a monotonic increase of metalinguistic awareness contingent on the proficiency of students in one, two, or three languages. Those highly proficient in three languages achieve the highest metalinguistic awareness scores. Furthermore, Lasagabaster (2001) examines the impact of students’ metalinguistic awareness in their first language (Basque or Spanish) on different skills in English. Interestingly enough, metalinguistic awareness positively influences English reading, grammar, writing, and speaking skills in a statistically significant way, though not listening skills.Footnote ¹⁹ It is one of the few studies where these skills are distinguished. Hofer and Jessner (Reference Hofer and Jessner2019) compare South Tyrolian (Northern Italy) primary school students in multilingual streams (where instruction proceeds in German and Italian) and traditional Italian-medium programmes and report better results of the former group in regard to proficiency in English and German as well as metalinguistic awareness (tested for Italian). They conclude that early multilingualism fosters language proficiency and metalinguistic awareness.

Again, we need to bear in mind that the above studies were carried out in contexts of additive multilingualism involving languages of high social prestige. Spellerberg (Reference Spellerberg2016) offers an interesting point of comparison in this respect (see also El Euch Reference El Euch2010). This study – placed in Denmark – involved 219 students aged 14–16 of which 106 were classified as monolingual and 113 as bi-/multilingual. The majority of the bi-/multilingual students came from immigrant families speaking a great variety of home languages. The study distinguished three language groups, namely monolingual Danish students, bi-/multilinguals who speak the majority language Danish at home, and bi-/multilingual who do not speak Danish at home (but one of the heritage languages, which are less prestigious in the context of Denmark). The measurement of metalinguistic awareness in this sample revealed a monotonic increase from the last group to the first. Monolingual Danish students scored highest.Footnote ²⁰ Moreover, metalinguistic awareness and school leaving exam results were shown to be strongly related. What needs to be borne in mind in the interpretation of these findings is that metalinguistic awareness also depends on socio-economic status and that the metalinguistic awareness test was administered in the majority language, Danish. In particular the latter fact may have skewed the results considerably (recall the discussion of Saer Reference Saer1923 in Chapter 1), as some heritage students may have found it difficult to complete the test. Nevertheless, the study serves as a warning that a simple equation of multilingualism, metalinguistic awareness, and proficiencies (academic achievement) falls short of reality.

2.3.3 Predictors of Metalinguistic Awareness

Given that metalinguistic awareness is a complex cognitive measure, it is influenced by a host of different parameters. Moreover, as metalinguistic awareness needs to be measured by some kind of test, the task design including its comprehensibility will determine the outcome. This is not different from other complex cognitive measures, such as intelligence or executive function.

It would appear intuitively plausible to claim that metalinguistic awareness is a function of the languages that one knows, but this cannot be more than a very rough estimate, since knowledge of language is difficult to define. For each language in a speaker’s repertoire, we need to determine age of onset, type of acquisition, length of exposure or instruction, the type of instruction (if applicable), the proficiency levels reached, frequency and domains of use, as well as the development of literacy. Some of these parameters predict language dominance (see Chapter 3). I consider explicit linguistic instruction an extremely important parameter, as it requires hard work to lay open the systemic properties of languages (see also De Angelis Reference De Angelis2007: 122–123). Uninstructed contemplation may lead to folk linguistic generalizations that perhaps reveal more about a speaker’s underlying stereotypes and cultural biases than metalinguistic awareness proper.

I wish to remain agnostic about which of the above parameters exerts the strongest effects, but it would appear that the influence of literacy should not be underestimated (De Angelis Reference De Angelis2007: 118–120). Literacy development practically forces the language user into developing metalinguistic awareness, as graphemes need to be related to sounds and meanings in a systematic way. It is usually taken for granted that phoneme awareness drives literacy, but there may just as well be a path in the opposite direction (Phillips, Clancy-Menchetti & Lonigan Reference Phillips, Clancy-Menchetti and Lonigan2008).Footnote ²¹

Finally, the kind of pedagogical approach used within the classroom matters for the development of metalinguistic awareness (see Chapter 4). Pedagogies explicitly addressing multilingualism and language diversity can be expected to give a boost to metalinguistic awareness. Allowing for the use of multiple languages in the (language) classroom rather than enforcing a monolingual habitat will be beneficial for the development of metalinguistic awareness. The traditional monolingual classroom valorizes and devalues languages according to their alleged classroom relevance and social prestige (see Fuller Reference Fuller and Hickey2020: 171 on “normative monolingualism”). It then becomes difficult to discuss and draw on the devalued languages, even though they could be helpful and relevant.

2.3.4 Summary

In comparison to executive function and cognitive development, the positive impact of multilingualism on metalinguistic awareness appears intuitively more plausible. Evidently, there is a closer relationship between the latter two concepts as they both relate to language proper. One may compare the development of metalinguistic awareness to that of meta-cultural awareness as a consequence of (conscious) travelling and cultural immersion. This also leads to the identification of similarities and differences between one’s own culture and the foreign culture as well as conscious reflection of it. If not done in an informed way, there looms the danger of forming cultural misconceptions (stereotypes) much as with folk linguistic generalizations.

There are several questions about metalinguistic awareness that remain unanswered by current research. First, it remains somewhat unclear whether bilingual and multilingual language users per se show higher levels of metalinguistic awareness in comparison to their monolingual counterparts. It appears evident that metalinguistic awareness increases in instructed language acquisition and perhaps during formal education in general. It is, however, less clear if multilingualism positively impacts metalinguistic awareness if there is no education system and hence no formal instruction. There is an important methodological problem here, as this would need to be measured with either very young children who did not receive formal education or in cultures without an education system. The very concept of metalinguistic awareness may be meaningless outside the education system.

Second, as argued above, monolingually raised children usually receive extensive grammatical tuition in their first language, starting already in elementary school, perhaps earlier. It can be taken for granted that this raises their metalinguistic awareness. However, if and to what extent this also facilitates subsequent language acquisition remains difficult to answer, but I suppose it does. In fact, even for bilingual children learning additional languages, the evidence is far from unanimous (see Rutgers & Evans Reference Rutgers and Evans2017).

Third, if it is the case that metalinguistic awareness facilitates the acquisition of additional languages, one may hypothesize that this also affects the scope and direction of cross-linguistic influence. For example, it could be the case that learners with higher levels of metalinguistic awareness show less cross-linguistic influence of the non-facilitating type, and perhaps higher degrees of facilitating influence. In addition, metalinguistic awareness may produce different effects of cross-linguistic influence in production and reception, speaking and writing. Such studies presuppose sophisticated research designs. Another problem concerns the specific domains of language learning where positive effects of metalinguistic awareness can be expected. The development of phonology and morphology may profit to a lesser (or perhaps higher) extent than lexical or syntactic development. The phonology of a foreign language is a well-known area of imperfect acquisition. Current research is suggestive of the fact that bilingual and multilingual learners indeed show certain advantages in this domain, perhaps as a consequence of heightened metalinguistic awareness. These issues will be followed up in Chapters 3 and 4.