The cognitive science of bilingualism has made important progress over the years, to which this survey of new research, translated from the Spanish edition by John Schwieter, makes a valuable contribution. Written for a general audience, investigators and students in the broader field of bilingual studies will profit greatly from the overview, bringing us all up to date on recent work. Reading the book reminded me that we need to keep a wide lens open on the bigger interdisciplinary picture, and on the eventual convergence among the parallel lines of work, from performance assessment in applied linguistics, neurolinguistics, and relevant theoretical modeling in evolutionarily psychology. With a passing reference to The Descent of Man on the primacy of speech, the five chapters concentrate on the first two lines of research.
Aptly, Chapter 1 begins with the first signs of the emergence of bilingual mental organization in early childhood. This is important because the evidence for early differentiation among the linguistic systems of children exposed to two or more languages frames much of the discussion on how the systems form in development. The defining indications start with the capability of all infants, bilingual and monolingual alike, to track phonotactic patterns in the speech stream, making possible (by eight months) the computation of regularities within words against the regularities between words. At the same time, infants’ sensitivity to the relevant features starts to zero in on the phonological contrasts specific to their mother tongue, losing an initial sensitivity to contrasts that do not appear in the language they are acquiring as first language (L1). The precocious spurt in lexicon building during the second year would not be possible without these attainments. Remarkably, in the case of simultaneous dual-language acquisition (2L1), the word segmentation settings for each system and the settings for phonological contrast for each, are deployed within the same developmental time frame, with “no significant delay” (p. 8). The potential interference from external contrasts, not helpful to the computations that bilingual acquisition carries out, are blocked. Then internally, except for transitional cross-language influence, the acquisition mechanisms also maintain the developmental “autonomy” of the two concurrently emerging L1 grammars. Early theories of an extended “fused” unitary system (an initial undifferentiated language network) have not been supported by findings from research on 2L1. The study of language acquisition when there are two interacting systems is important because it provides evidence for the different proposals about “… the linguistic knowledge that babies have from an early age” (p. 5). Chapter 1 summarizes the interesting results from studies of four-month-old Spanish–Catalan bilinguals’ ability to discriminate between the two languages. Regarding a hypothesis of evolutionary antecedents of this ability, on page 11 we consider the possibility of homologous sound discrimination capabilities in other primate species.
Chapter 2 introduces the important topic of cortical representation, of the distributed neural circuits that correspond to knowledge of two languages. An important aspect of the description that research will converge upon is that of interactivity (p. 30), and how learning comes to be subserved by new sets of connections. But so far it has been the study of brain trauma, of different kinds, that has provided more tractable evidence, over time contributing to the rich discussion among theorists. Here a key comparison is between associated deficits and dissociated deficits. In this way, learning and loss provide data on the same problem. How might cognitive functions be represented by shared domains and/or by domains that are specialized in some way? In the end, the relation of “and/or” might be the idea to keep in mind that accounts for all findings. The chapter makes reference to “systems” that can be thought of in terms of “circuits”, even though the figure of “regions” should not be discounted either, reminding the reader that we are still at the beginning, despite the advances that have been made so far. Within “language” might there be representations/domains specialized for components and subcomponents of the grammar, possibility suggested by dissociations that patients exhibit (pp. 34–35)? At the same time, associated (or parallel) impairments also suggest how some representations/domains are interdependent and not domain-specific. In the end, the evidence from language loss will have to be integrated with the evidence from learning.
In comparing impairment to each of the language systems in bilinguals, what conclusion should we draw from the fact that the parallel variant is the most common? One avenue of investigation that this finding points us toward is to ask the question about domains that are “shared”, or that “overlap”. Parallel functioning implies that “the areas involved in the representation and processing of a bilingual’s two languages are the same” (p. 42). The same overall principles are applied to both language systems and “the properties that are important for language organization in the brain are the same for [the] two languages” (p. 39). Thus, within one and the same language system, substructures might account for differentiation and specialization. In any case, the possible existence of language-specific, “independent”, substructures, doesn’t in any way preclude massive interactivity.
Chapter 3 follows up on the discussion in the previous chapter on how first language and second language (L2) are represented, the key distinction to keep in mind being the difference between:
° processing components (e.g., task-specific interface mechanisms), and
° competence, or knowledge, representations.
Even though this distinction has not been definitively established in the cognitive sciences across all domains, the interpretation of findings in the field at least suggests that it is psychologically real in some way. One example is from the study of the bilingual lexicon. Which aspects of vocabulary development in bilingual children, revealed in different kinds of language ability (e.g., related to literacy learning), reflect actual knowledge of words, and which reflect instances of processing or on-line access? The discussion of related studies in Chapters 3 and 4 underlines the importance of controlling for intervening and confounding performance factors. These are, hypothetically, cross-language interaction (CLI) (e.g., interference) and other processing mechanism effects that involve the interaction between language and domain-general cognition. On naming tasks, even highly proficient bilinguals evidence a less efficient access to the lexicon (slower with more frequent errors) than monolinguals, even when tested in their dominant language (pp. 67–69). The finding of less reliable naming (from illustrations) appears to be related to falling into tip-of-the tongue states more often than monolinguals in controlled experiments, as well as lower scores on verbal fluency (rapid naming of category members). These results clearly illustrate the effects of CLI.
Inhibitory mechanisms make possible the very use of one language or another, systematic alternation (as in translation tasks), and grammatical and coherent code-switching. Inhibition also varies depending on task demands. Obviously, CLI is not blocked during code-switching; but, interestingly, when bilinguals perform listening and speaking tasks in a “monolingual mode”, interference from the non-selected language cannot be avoided completely. But again, without control over inhibition, fluent and comprehensible verbal expression on the part of bilinguals would be impossible. Comprehension would be affected in a different but analogous way. Thus, one way that researchers describe this phenomenon is that access to lexical items for bilinguals is more costly and “inefficient” as a result of the continuous controlling of potential interference. For other tasks the additional costs are incurred in recovering from inhibition (pp. 70–71). Importantly, these patterns of dual-language performance appear more clearly in the difference between dominant L1 and non-native L2 bilingualism (as opposed to equal proficiency). One explanation for the disparity is that later-acquired L2 might call upon control processes and deliberate learning strategies to a greater degree than L1 that relies more on implicit and automatic mechanisms. We return in Chapter 5 to the distinction between reflective/deliberate learning and intuitive/automatic acquisition in the discussion of System 2 and System 1 capabilities. The concept is important in contrasting the bilingualism of late-L2 learning and early simultaneous bilingualism.
What is evident from all these examples of demand on processing is that they are in principle independent of the language competence representations of the bilingual’s language systems. A possible qualification of this hypothesis might be that these processing costs and surface incursions come to affect the actual competence components of the weaker, non-dominant, language in imbalanced bilingualism, leaving those of the dominant language system completely intact. This possibility is presented implicitly in the section on language attrition (pp. 61–65). Over time, the cumulative effect of language use and processing factors such as inhibition, interference, and imbalances in ease of access might play a role in the erosion of actual competence in the weaker language system.
From this overview, the various threads from the relevant studies are gathered together in one of the important themes of the book: that researchers and practitioners (especially language teachers) need to carefully evaluate the many claims regarding advantages or disadvantages of bilingualism in the realm of language ability, especially when the effects are presented as significant and wide-ranging. In particular, the strongest claims, favoring one side or the other, have not been confirmed under close scrutiny.
Chapter 4 follows up on studies that focus on the related question of the potential effects of bilingualism on general cognition. Some of these have already been examined in the previous chapters because domain-general capacities and processes participate in all aspects of language ability and language use. A good example is the comparison between monolinguals and bilinguals on Theory of Mind problems (pp. 83–87). Why would monolingual children take longer (according to studies) to attain the successive milestones of this aspect of metacognition?
The attentional system, or executive control, hypothetically, would play a role in regulating interaction between the language systems: How high, so to speak, from one context of use to another should the filter be set on the inhibition of competing activations? On the one hand, we can attempt to measure these processes on-line. Then on the other hand, there is the question of whether greater ongoing involvement of the attentional system has an effect on the steady state level of functional architecture: “… to what extent bilingualism can imply an additional exercise that results in more efficient executive functioning” (pp. 95–97). Other examples, sometimes referred to as “higher-order”, are the ability to focus attention on relevant and irrelevant information for resolving conflict or contradiction, and performance on tasks that require switching classification criteria. Indeed, some of the reports for an advantage for bilinguals come from studies with young infants, months before they utter their first word, the proposed factor being related to early development of flexibility of the attentional system (Kovács & Mehler, Reference Kovács and Mehler2009; Sebastián-Gallés, Albareda-Castellot, Weikum, & Werker, Reference Sebastián-Gallés, Albareda-Castellot, Weikum and Werker2012). Hypothetically, in these studies children inhibit information from the initial trial and then revise predictions in accord with new task demands (p. 104).
To recap, and to elaborate somewhat upon, the author’s theme: on balance it appears that in the end the weaker hypotheses regarding the positive effects on higher-order cognitive functions are more likely to prosper; and for each domain the evaluation will need to be made case-by-case. In addition, the same cautious approach will probably allow us to discard all hypotheses that predict deficits in language and intellectual capabilities; i.e., in the actual knowledge/competence representations. In the end, findings of lower performance on vocabulary tests by bilingual children, for example, could be easily explained by processing factors and the complementary distribution of lexical items between Language A and Language B lexicons. Theories that certain kinds of imbalanced bilingualism (the so-called “subtractive” variant, prelude to L1 attrition) result in an ill-defined “semilingualism” have now been decisively discredited. Deficient (non-native) competence in both language systems only appears in abnormal language development, the same Specific Language Impairment condition that afflicts monolinguals.
Taking the liberty of drawing a tentative conclusion from the studies presented and analyzed, the following is a proposal for future research: that there is no advantage of bilingualism per se within the linguistic domain, in the narrow sense, because the critical evidence for it would appear in early childhood simultaneous dual-language acquisition. It happens that the research comparing monolingual and bilingual child language acquisition shows that the major developmental milestones are attained on the same timecourse. Neither bilinguals nor monolinguals surpass the other qualitatively.
Then, possible advantageous effects of bilingualism in the domain of general cognition would arise from language use of two languages that implies one or another type of higher-order processing challenge, deployment of cognitively demanding operations to some degree, different from language use that is of the implicit, automatic, and System 1 type. This hypothesis implies that monolinguals immersed in the cognitively demanding, System 2 type, use of language (for example in early advanced literacy and sustained experience in the comprehension of complex oral tradition discourse, mental construction and performance of higher-order narrative, etc.) may score within the same range as bilinguals on the assessments described in first four chapters of the book. The cognitive challenges of second and third language learning and cognitively demanding bilingual communicative tasks present children with one opportunity at hand (among others), one that happens to be widely available to them in multilingual speech communities. For example, studies have suggested that enhanced conceptual development may be associated with the experience of second language learning as opposed to bilingualism per se (Le Pichon, de Swart, Vorstman, & van den Bergh, Reference Le Pichon, Swart, Vorstman and van den Bergh2010). The cited study compared early simultaneous bilinguals and children who became bilingual via L2 learning.
Chapter 5 delves into the study of the interfaces between language (in the case of bilingualism) and other cognitive systems, focusing on affect and decision-making. Applied issues in the social realm come forward in the studies that we consider in this field.
All of the above takes us back to the research question presented in Chapter 2, to now serve as a conclusion to this review. To his credit, the author mentions that his emphasis on “parallel impairment”, and his interest in a model that takes this tendency into account, is controversial; but it’s not clear in what way it is, especially or necessarily. To great benefit for the reader, the chapter cites counter-examples, of dissociation, that appear to favor a model based on cortical structures that are specific to each language system. So, by all accounts, including the balanced presentation in this book, we can see that the different empirical questions in play need to be kept open for continued discussion. Then all the relevant examples and counter-examples will need to be reconciled.
Up until now, in the book and in this review, we have made reference to the language “systems”. But if the areas involved in the representation and processing of the bilingual’s two languages are the same, as are the fundamental properties of language organization, then it might be better to think of Language A and Language B as subsystems that form part of one single linguistic system (Paradis, Reference Paradis2004). As the section in Chapter 2 (pp. 40–49) on how a L2 is represented differently from the dominant mother-tongue L1 pointed out, L2 ability may recruit more extensively from cognitive-general domains than the L1. But this turns out not to be different in all respects from L1 ability, because for challenging higher-order language tasks L1 performance also recruits from the same extralinguistic capabilities. Nevertheless, within the same linguistic system, the evidence suggests that L1 and L2 form two subsystem networks, subject, as in all cases of dual-language competence, to CLI, in both directions. Here, research will need to specify what we mean by competence representations of L1 and L2.
The same will need to be specified in the case of 2L1, where balance from early simultaneous acquisition is maintained and neither is the dominant language. One proposal, that applies also to L1 and L2, favors a differentiated subsystems model in contrast to an integrated model; see Kovelman, Baker, and Petitto (Reference Kovelman, Baker and Petitto2008) for discussion. In addition, there are different versions of each proposal that research will need to sort out for the pending difficult questions. For example, how do language-specific domains interface, internally, within each language subsystem of the bilingual system, then cross-linguistically to corresponding nodes and networks in the other language subsystem? Then, how do they all interact with general cognition (including conceptual structure)?
Aside from building a model that accounts for all patterns of impairment and recovery, findings from a number of bilingual performance contexts will be informative:
° the remarkable early separation in the processing of Language A and Language B by infants,
° followed soon after by awareness of and ability to discriminate between the language subsystems,
° grammatically constrained switching in bilingual speech by children,
° inhibition of the non-selected language (subsystem) in skilled translation and other interference-prone tasks,
° how in attrition of A or B the content of the dominant language representation is “closed off”, blocking interference from the eroding weaker language.
Then, on the other hand, how is the unitary conceptual system accessed, assuming that in bilinguals it is shared between the language networks, and not differentiated?
These and other related questions await new research opportunities that this concise update on the Science of Language lays out with an upbeat and optimistic forecast. The way the outstanding controversies are presented is helpful as it points to a way forward for examining the current evidence based on what we have learned so far, especially, after many years, regarding the converging conceptions already achieved in the field.
