It is immensely difficult to understand how children learn one language.
The competing descriptions each explain only part of the process, so a complete understanding requires integrating pieces from different per-spectives. The need to weave together a larger fabric from distinct frag-ments is not surprising given the complexity of the enterprise and the number of factors that impinge on children’s experience in language learn-ing. This complexity is multiplied when one considers the factors relevant to a child’s experience in learning two languages.
The questions that one need ask about children learning two languages are different from those guiding the study of monolingual language acqui-sition but they reflect the same divide between formal and functional paradigms. For example, the difference between simultaneous and se-quential acquisition of a second language matters only under certain inter-pretations of the process of first-language acquisition. Applying formalist assumptions, both varieties of second-language acquisition should be sim-ilar because they are guided by the mechanism of the language acquisition device and shaped by the constraints of universal grammar. Hence, any input to the child would lead to the construction of the appropriate gram-mar. Following functionalist assumptions, however, the sequence and tim-ing of second-language acquisition would be different dependtim-ing on whether the two languages were acquired at the same time or in sequence.
In these views, social interaction and previous knowledge drive language acquisition, placing a significant divide between the experiences associ-ated with simultaneous or sequential acquisition of a second language.
A fundamental question about children learning two languages is whether the acquisition of each language resembles the process and path experienced by a monolingual child learning only that language. How-ever, the apparent simplicity of this question is deceptive, as it
incorpo-rates a set of assumptions about the nature of language, the process of acquisition, and the structure of mind. Again, whether or not we consider that children acquiring two languages are comparable to monolinguals is rooted in a position regarding the nature and acquisition of a first language.
A plausible but complex view of language acquisition is achieved by incorporating aspects of both the acquisition of linguistic competence explained by formal theories and the acquisition of communicative com-petence explained by functional theories. This synthesis of perspectives seems more imperative as a means explicating children’s acquisition of two languages than it does for the usual monolingual language acquisi-tion. As again becomes apparent in the discussion of the cognitive impact of bilingualism (Chapter 7), the study of bilingual children advances the-orizing by forcing explanations onto a broader palette. Similarly, ques-tions about children’s language acquisition allow greater access to some of the central issues of mind when children are learning two languages than when they are learning just one. This point is discussed in more detail below.
The integrative view that follows from including both formal and func-tional perspectives appears complex – it cannot be stated in a few simple assumptions and constrained by a limited number of factors. One conse-quence of this complexity is that it removes the possibility of generating simple dichotomies and describing concepts categorically. Hence, the differences between first- and second-language acquisition, between si-multaneous and successive acquisition, and between presence and absence of proficiency in a specific language need to be examined more analytically and described in greater detail.
The entire process of using language is different when two languages are available. A dichotomy in which mastery is assessed individually for each of the two languages fails to capture the essential nature of having two languages at one’s disposal. Grosjean (1989, 1998) proposes a solu-tion in which a scale is used to describe language competence for bilingual speakers. He postulates that bilinguals engage in a range of language functions each of which varies on a dimension from monolingual mode, mimicking the language behavior of a monolingual speaker of that guage, to bilingual mode, naturally incorporating aspects of both lan-guages into production. In a bilingual mode, for example, language mix-ing and switchmix-ing is normal and comprehensible. Bilmix-inguals move freely along this dimension in response to social and linguistic circumstances.
He calls this the bilingual view, in contrast to the monolingual view that
would ascribe two language competencies on this dimension to a bi-lingual, each of which corresponded to a monolingual speaker. The under-lying linguistic competence that permits this movement along the dimen-sion from bilingual to monolingual modes indicates that linguistic representation for bilinguals is importantly different from that of mono-linguals. Therefore, we should not expect language acquisition in each of the bilingual child’s two languages to replicate exactly the pattern experi-enced by a monolingual child learning only one of those languages be-cause the representational system for both languages are different.
Another question is whether a second language must be learned by a certain age in order to achieve native-like competence. If there is one certainty that people hold about language acquisition, it is that children can learn a second language better than adults can. Indeed, there is no shortage of anecdotal and empirical evidence to support this supposition.
A frequent interpretation of this prodigious talent by young people is that it is the reflection of a critical period for language acquisition in humans.
The claim has serious implications because of the logical connection be-tween first- and second-language acquisition. If it could be demonstrated unequivocally that there is a critical period for first-language acquisition (no small feat given the nature of the data required for such a proof ), then it may or may not follow that humans are constrained in their ability to learn another language later in life. In other words, establishing the exis-tence of a critical period for first-language acquisition would tell us noth-ing about whether or not such a constraint exists for second-language acquisition. However, if there is a maturational limit on the time period during which a second language must be learned, then it follows neces-sarily that any language learning by humans is so constrained. This is a powerful generalization about human cognition and has implications for issues such as how languages are learned, the status of a linguistic biopro-gram, and the organization of mind to include a contained language module. These issues resonate profoundly through our conceptions of intelligence and the human potential for learning. Access to these issues, then, begins by deciding whether there is a critical period for second-language acquisition.
Mastering the System
We tend to think of children’s minds in protective terms: some things are too hard for them to understand, some details are unnecessarily confus-ing, and some information is too burdensome. These patronizing cautions
emanate from the assumption that the mind is a container – a vessel of limited capacity that must be filled with care. It follows, surely, that trying to squeeze two complex languages into such a fragile receptacle would lead to bewilderment, frustration, and failure. However, nothing we know about memory substantiates these fears. Indeed, the fact that millions of children routinely grow up with more than one language in their environ-ment and appear to suffer no obvious trauma should allay the concerns of most parents.
Early research on the possibility of children learning two languages to an equal and acceptable level of proficiency was far less sanguine in its conclusions. Macnamara (1966) undertook a detailed review of seventy-seven studies published between 1918 and 1962, performing a sort of early “meta-analysis.” These studies included numerous languages and language pairs as well as various situations and circumstances for chil-dren’s bilingualism. Although he gave few methodological details for the studies, he did point out which ones were more trustworthy and in fact excluded some studies where he believed that methodological flaws under-mined their conclusions. Linguistic assessments included tests of vocabu-lary, spelling, reading, and grammatical complexity, to name a few. Mac-namara concluded that “bilinguals have a weaker grasp of language than monoglots” (p. 31), and supported this conclusion by listing a running score for the number of studies in each linguistic skill that reported each conclusion. For example, twenty-four studies investigated vocabulary, of which six reported bilinguals were equal to monolinguals and eighteen reported vocabulary superiority for monolinguals. Macnamara thus con-cluded that bilinguals had vocabulary deficits. Six studies investigated spelling, of which one reported a bilingual deficit and five showed no difference. Thus it was concluded that spelling is not affected by bilingualism.
Macnamara offered four reasons for his conclusion that bilingualism causes language deficits. First, linguistic contrast creates interference, es-pecially for highly divergent languages (cf. Lado, 1957); second, cultural assimilation is crucial to language learning and is not usually present for bilingual children (cf. Whorf, 1956); third, language models are often inadequate (parents may not be native speakers) and opportunities to learn, such as school attendance, may be absent; and fourth, time available to learn each language is less for bilingual children, including as well the effects of age, ability, and incentive (cf. critical period hypothesis reviewed below). He concluded that the time factor accounts for the deficits in vocabulary, that linguistic interference and faulty models account for
deficits in grammar and phonetics, but that the real explanation for the deficit would require complex interactions among all of these.
Macnamara’s conclusions that bilingual children failed to learn both languages adequately have not been supported by more recent research and thinking about these issues (discussed below). Aside from meth-odological flaws in the research (some of which Macnamara details), there is an illogic to the argument. Consider a paradigm case in which a child learns Spanish at home as the family language and attends school in English. The community is primarily Spanish-speaking and the child has only school as an environment to develop English skills. It is not surpris-ing that the type of competence that the child develops in each of these languages is different. In an ideal situation, we would expect the child’s spoken Spanish to be equivalent to Spanish monolinguals and the child’s academic English to be equivalent to English monolinguals, but social circumstances rule against that outcome. The child’s oral Spanish may be compromised by such factors as dialect, family educational levels, and range of communication experiences. Similarly, academic English might be diminished by the lack of support at home for high levels of English usage. The point is that assessing children’s linguistic skill requires under-standing children’s language experiences, and the objective outcomes of such assessments are uninterpretable without knowing about the context.
Invariably, the language assessments used in these studies were formal tests administered in both languages. There is no reason to believe that children should have the same type of competence in both languages given these vastly different experiences with each. Romaine (1999, p. 259) makes the same point: “It does not make much sense to assess bilinguals as if they were two monolinguals since it is unlikely that a bilingual will have the same experiences in both languages.” In short, these studies may tell us very much about the social context for learning language but almost nothing about the nature or impact of bilingualism on the language profi-ciency that children can achieve.
There is some empirical evidence that the structure of language profi-ciency is not the same in the two languages of a bilingual. Lemmon and Goggin (1989) tested a large number of university students who were Spanish-English bilinguals and found a complex pattern of language pro-ficiency. Using a battery of language tests, they found variable proficiency for the different skills and different factor analysis structure for English and Spanish. The lesson is that we require more sensitive tests of language proficiency to understand a learner’s competence in each language. Gener-alizations across types of proficiency must be made cautiously.
Recent research has been more careful in comparing abilities and at-tending to the contexts in which children learn different languages. Not surprisingly, the results are quite different from those using less sensitive methods. The attempt is to carry out a detailed study of linguistic skills when the monolingual and bilingual groups are known to be equivalent in other relevant factors. In this way, the process of learning each aspect of linguistic competence is traced individually for the various language skills, and each one can then be compared with the progress made by mono-lingual children learning the same system. These studies not only inform our knowledge of the effects of bilingualism on children’s language acqui-sition, but also contribute to our understanding of how these individual aspects of linguistic competence develop and interact.
Two questions that are fundamental to interpretations of the nature of language and its acquisition are, first, the relation between language and the conceptual system (alternatively construed as the relation between language and thought), and, second, the possibility of an innate mental structure for language. These questions address two of the three issues that distinguished between formal and functional theories of language acquisition described in Chapter 2. The first is the problem of the auton-omy of language from other cognitive processes and the second is the role of innate structure in language acquisition.
These questions are staggeringly difficult to investigate when children learn one language and have one conceptual system. Bilingual children, however, offer an opportunity to disentangle some of their important dimensions. For the first question, we can examine the way that bilingual children acquire vocabulary in their two languages; for the second, we can study the syntactic and phonological systems of young bilingual children.
These examples will provide descriptive information about the language development of bilingual children and comparative possibilities to assess the generality of the acquisition by monolingual children outlined in Chapter 2, and offer a window into some of these more profound ques-tions about the nature of language and thought. We begin by examining the progress of acquisition for children learning two languages, then con-sider the evidence for a critical period for language acquisition, and finally reflect on the implications of the story for these more basic issues in language and cognition.
Lexicon
How does the need to learn two vocabularies influence children’s pro-gress in language acquisition? Do bilingual children share their processing
and acquisition resources across the two languages, essentially controlling half the vocabulary range in each language that would be mastered by a monolingual child? Alternatively, do processing resources expand to fill the considerable needs required to store two lexical systems? Both posi-tions seem unlikely. Children’s developing cognitive capacities surely im-pose limitations on the breadth of information that can be stored in accessible memory, ruling out the possibility of two complete vocabul-aries. At the same time, however, children indeed seem capable of learning more than the minimal range of vocabulary required for one functioning language system. Somewhere between these options, bilingual children probably function in an interlingual purgatory, less masterful than a monolingual in either language but surely more extensive in their com-municative possibilities than any monolingual.
As with many aspects of language acquisition, hypotheses such as these defy empirical examination. If there is one prevailing certainty about vocabulary acquisition, it is its variability across children. In fact, the more normative data that are gathered, the more variability emerges as the predominant characteristic (see, e.g., Fenson et al., 1994). A properly controlled study of the effects of bilingualism on lexical acquisition would need to compare the vocabulary of bilingual children in terms of its size, conceptual coverage, and linguistic sophistication to the vocabulary they would have had in each language had they been learning only one of them.
Such hypothetical speculation is quite obviously untestable, so com-parisons must be waged across groups and, therefore, between individ-uals. The inevitable differences in ability and circumstance between any two children make these contrasts at best a weak proxy for the issue of real interest, the influence of bilingualism on vocabulary growth.
This methodological problem of comparison does not arise to the same extent for other aspects of language, such as syntax or phonology, proba-bly because in these cases the sheer quantity of information learned is not the most significant variable. In syntactic acquisition, for example, it is sufficient to establish the progress children make in terms of utterance length, type of error, and structural sophistication. Although there is cer-tainly individual variation in these dimensions, it is reasonable to assess progress according to age norms, and the balance between the two lan-guages is more stable than the possibilities for enormous asynchrony that exist for vocabulary.
In addition to being a problem in its own right, the difficulty of measur-ing and interpretmeasur-ing vocabulary size in the two languages of a bilmeasur-ingual child also emerges simply from the complications of research into
chil-dren’s bilingualism. Research comparing bilinguals to monolinguals nec-essarily uses between-subjects designs. Such research always requires ex-tra precaution in assuring that the variable of interest is the only relevant difference between the subjects in the two groups. For comparisons be-tween bilinguals and monolinguals, this is not simple to arrange.
Researchers conducting these studies attempt to create groups that are comparable in all respects but their bilingualism. Partly, the assumption of equivalence is a convenient fiction because children growing up with two languages are necessarily different from children growing up with only one in ways that are not revealed by assessments of language ability (see discussion in Chapter 1). Nonetheless, such simplifying assumptions are essential if progress is to be made in understanding the nature of develop-ment for bilingual children. To this end, a test of receptive vocabulary, such as the Peabody Picture Vocabulary Test (PPVT) is sometimes used both as an indication of the degree of balance between the two languages and as an index of the comparability of the bilingual children’s language abilities in each language to those of monolinguals. Ideally, it should be shown that the receptive vocabularies for the monolinguals and bilinguals in their common language are similar. It is further beneficial if the recep-tive vocabularies in the bilingual’s two languages are comparable to each other. This would support the claim that the child is a relatively balanced bilingual, a representative of an idealized class of bilinguals.
Published research generally meets the selection criteria of guarantee-ing the comparability of the monolguarantee-ingual and bilguarantee-ingual children studied;
indeed, they would be deemed unworthy of publication if they did not.
Nonetheless, studies frequently report differences in the PPVT scores of monolinguals and bilinguals (Ben-Zeev, 1977; Bialystok, 1988; Merriman
& Kutlesic, 1993; Rosenblum & Pinker, 1983; Umbel et al., 1992). Some research has used the differences between PPVT scores in the two
& Kutlesic, 1993; Rosenblum & Pinker, 1983; Umbel et al., 1992). Some research has used the differences between PPVT scores in the two