Looking to atypical populations to illuminate the functioning of core cognitive systems has a long history in the behavioral sciences (e.g., Caramazza, Reference Caramazza1986; McCloskey, Reference McCloskey1992). Pierce, Genesee, Delcenserie, and Morgan (Reference Pierce, Genesee, Delcenserie and Morgan2017) continue that tradition in their keynote article, here by shedding light on how the development of phonological working memory (PWM) may be shaped not only by individual genetic variation but also by various features of the language acquisition context, whether that context is enriched, as in bilingualism; delayed, as with cochlear implantation (CI) or international adoption (IA); or impoverished, as with deaf children of hearing parents or repeat cases of otitis media. Their review of the literature demonstrates that a sensitive (/critical) period for phonological development, encompassing variation in quality, quantity, and timing of early exposure, has crucial consequences for an individual's level of PWM attainment and, by a compelling extension, for aspects of ongoing language development such as vocabulary learning and reading skills.

Yet the mechanisms that drive the differences in PWM across those groups may be distinct. The authors see this, and they also acknowledge that the path from “enriched” linguistic input to enhanced PWM may be indirect, perhaps passing through other executive functions (EFs) such as cognitive flexibility and inhibitory control. These same EFs, moreover, have been found to be correspondingly weak in the CI (e.g., Figueras, Edwards, & Langdon, Reference Figueras, Edwards and Langdon2008) and IA (e.g., Hostinar, Stellern, Schaefer, Carlson, & Gunnar, 2008) populations. With EFs as with PWM, an enriched acquisition environment leads to stronger cognitive functioning, while an impoverished acquisition environment leads to weaker cognitive functioning. Thus, many of these contrasting nonlinguistic effects follow from the spectrum of early experiences reflected in the keynote that the question naturally follows: is it linguistic representations (and, more specifically, phonological representations) that drive these groups’ respective PWM advantages and deficits, or is it instead other, more basic features of cognitive development? In this commentary, I would like to suggest one possible approach to teasing phonological representations apart from EFs, including the target PWM. I first sketch an argument for keeping bilingualism on the continuum with the impoverished and delayed exposure contexts, rather than casting off the reported superior bilingual PWM ability into the admittedly murky waters of the bilingual EF advantage. I introduce next some additional evidence to propose that, even if an explanation for the bilingual PWM advantage does not need to rely on enhanced bilingual EF more generally, there may be good reason to continue to explore how EFs might fit in the account. Specifically, while several intersecting fields are currently enlivened by a healthy disagreement about the oft-presumed bilingual advantage, there appears to be far less disagreement that EFs, and many skills dependent on them, are sorely disadvantaged in children in the other examined contexts. If it is possible to identify a route of positive influence between PWM and EFs, then along the keynote's continuum is where it might be found. Below I gesture at another documented bilingual advantage, in the sociocognitive domain, that may illustrate the promise of this approach.

A bilingual is a person who can speak two languages, which naturally presupposes significant experience with “enriched” linguistic input. Pierce et al. (Reference Pierce, Genesee, Delcenserie and Morgan2017) highlight simultaneous and successive bilingualism, noting at first that both groups could be at a disadvantage because of a reduced quantity of input in each language, and that successive bilinguals in particular could be challenged by their exposure to the second language after the sensitive period has closed. However, they rightly report that long-term language outcomes are generally favorable in this population and that bilinguals’ language experiences often lead to EF enhancements (cf. Bialystok, Craik, & Luk, Reference Bialystok, Craik and Luk2012; although see de Bruin, Treccani, & Della Sala, Reference de Bruin, Treccani and Sala2015). How does the mind cope when confronted with two phonologically, lexically, and syntactically distinct linguistic systems? Evidence from infant research (e.g., Bosch & Sebastián-Gallés, Reference Bosch and Sebastián-Gallés2003), functional neuroimaging (e.g., Abutalebi, Cappa, & Perani, Reference Abutalebi, Cappa, Perani, Krool and de Groot2005), and longitudinal corpus data (e.g., Hsin, Reference Hsin, Côté and Mathieu2014; Hsin & Legendre, Reference Hsin and Legendre2017) suggests that, to greatly oversimplify the lesson, simultaneous bilinguals’ two languages are closely integrated early in life and are distinguished only when (in development) and where (in the system of representations) a distinction is necessary.

Given the relatively constrained syntactic and semantic structures found in the languages of the world (and in particular in the structures used by and with children), primarily that distinction is imperative in two domains: the phonology and the lexicon. It seems to me that these two levels of linguistic representation could conspire to support the PWM gains found in bilingual children, both simultaneous and successive, without recourse to more opaque and complex constructs like response inhibition. The reliable course of phonological development in monolingual infants is reproduced at a slightly slower pace in bilinguals (cf. Bosch & Sebastián-Gallés, Reference Bosch and Sebastián-Gallés2003), given the larger set of representations the latter group must master. However, any negative impact (e.g., of having less repetition of lexical items or of hearing fewer words per language in total) that a bilingual child might experience may well be made up for by the metalinguistic awareness and complex conceptual connections she develops as her lexicon grows. For example, young bilingual children mostly avoid using a disambiguation heuristic to associate a novel noun with a novel object rather than a known one (Byers-Heinlein & Werker, Reference Byers-Heinlein and Werker2009), unlike monolingual children, because bilingual children's linguistic experience teaches them that an object can have more than one name (i.e., one per language). Nonnegligible overlaps between a bilingual child's two lexicons plausibly lead to stronger semantic representations for overlapping items (assuming a generic language processing and storage model), which in turn would support further vocabulary acquisition, vocabulary that, as the authors would likely have it, could serve to maintain a facility with PWM.Footnote ¹ Core EFs can be omitted from the explanation.

Leaving core EFs out of the bilingual PWM-lexicon pipeline does not mean they will not or should not show up somewhere else. Another cognitive domain in which children appear to benefit from bilingualism is in the development of theory of mind (Goetz, Reference Goetz2003; Kovács, Reference Kovács2009), that is, in their ability to take another person as an intentional agent with thoughts and beliefs potentially distinct from their own. A leading account of theory of mind implicates EFs in arguing, very roughly, that tracking others’ mental states is not as developmentally sophisticated or as difficult as is inhibiting one's own beliefs (e.g., Carlson & Moses, Reference Carlson and Moses2001). It is often the case that the bilingual advantage for preschoolers on theory of mind tasks only emerges when studies control for language ability (e.g., Nguyen & Astington, Reference Nguyen and Astington2014): bilingual children's proficiency in the language of testing may be weaker than monolinguals’, but they compensate for it with other stronger skills (e.g., EFs and social awareness). That is, bilinguals’ theory of mind advantage may appear in spite of their progress in grammatical development rather than because of it. Might an analogous compensatory mechanism be found in any of the other populations for whom PWM is at risk? Deaf children with CIs frequently lag behind their hearing peers in theory of mind development even when language is controlled for (e.g., Ketelaar, Rieffe, Wiefferink, & Frijns, Reference Ketelaar, Rieffe, Wiefferink and Frijns2012; Peterson, Reference Peterson2004). In addition, while it is clear from the keynote that IA children display relatively weak PWM, an additional demographic split can inform the inquiry further: IA children who had been institutionalized before their adoption into US homes have weaker verbal skills than those who had resided in foster homes in their birth country, and they have weaker theory of mind ability as well (Tarullo, Bruce, & Gunnar, Reference Tarullo, Bruce and Gunnar2007). Controlling for the linguistic difference, formerly institutionalized children still show worse false-belief performance than their peers who had spent their first months in foster care (cf. Colvert et al., Reference Colvert, Rutter, Beckett, Castle, Groothues, Hawkins and Sonuga-Barke2008), suggesting a deeper EF or social deficit.

If theory of mind relies on EFs and varies systematically between groups that differ in more than just the quality of their early language exposure, then it may be advisable to look back to the situational factors that characterized those groups in the first place. In other words, the observed effects of delayed or impoverished linguistic input may have been confounded thus far by other kinds of impoverishment. What we know about the generational persistence of adverse developmental circumstances indicates that disrupting the connection between weak PWM and weak language outcomes would require rich theory and ingenious intervention, targeting not only PWM and other aspects of oral language but also their academic counterparts like literacy, so we might now look to the wider array of benefits that bilingualism can confer, and how it may do so, to help guide that inquiry. Pierce et al. (Reference Pierce, Genesee, Delcenserie and Morgan2017) have succeeded in harnessing diverse and atypical language acquisition contexts to explore how early experience constrains PWM development in all acquisition contexts, including monolingual ones. In so doing, they have not only proposed a novel continuum of language acquisition environments and an appealing causal account of how early experiences shape later performance but also, perhaps inadvertently, recommended a multifaceted reevaluation of some of the very bilingual advantages that they invoke.