Maturational Constraints in Language One and
Language Two: A Second Look at the Research on
Critical Periods

Norbert Francis
Northern Arizona University

Abstract

Research on first language acquisition in young children and second language learning in older children and adults has examined the possible role played by maturational constraints (related to the notion of a critical period for language acquisition). The following article reviews findings from investigations that studied the special circumstances of delayed first language learning and the effect of age on second language learning. Findings suggest that: (a) the observed phenomenon of a critical period in language development merits further study, and (b) the ability to attain native-speaker grammatical competence may actually begin to diminish earlier than puberty. The discussion of the current research presents bilingual educators with a new framework from which to reconsider a number of pedagogical issues and recent controversies regarding school language policy.

Sections of the Article

Introduction

The following review and discussion of the literature will propose a focus for future research in an area of the study of bilingualism that has recently received relatively little attention: maturational constraints on first language (L1) acquisition and second language (L2) learning. Also referred to as critical (or sensitive) periods, the discussion of maturational factors has most often been associated with theories of child language development that have emerged from the Universal Grammar/Transformational-generative school (UG for short). Indeed, in recent years within the field of bilingual education, UG approaches seem to have largely fallen into disfavor or have come to be viewed as not pertinent to the problems of dual language teaching. A survey of the current teacher preparation literature reflects this apparent trend; texts often make limited reference to concepts and findings explained by UG investigators (on occasion strongly qualified to distance the author's interpretation of the findings from the theory in general), make no reference to UG hypotheses, or explicitly challenge them (see Lessow-Hurley, 2000; Faltis, 2001; Genesee, 1994; Hernández, 1997; Pérez-Torres & Guzmán, 1996; Cummins, 1996; DeGaetano et al., 1998; Durkin, 1995; Piper, 1998). Clearly, the present stage of inquiry and investigation does not warrant a convergence around any of the competing theories, and researchers working within one or another UG framework would be the first to warn against any far-reaching pedagogical applications of their findings, which by and large remain restricted to theoretical domains. One domain of the study of language development, however, does appear to hold out great promise: resolution of the key research questions related to critical periods in L1, and optimal age for second-language learning will decisively inform a series of current language policy and language-teaching debates.

For now, a discussion of these questions among bilingual educators will help place the debates into a broader perspective, the primary immediate benefit being a more reflective assessment of the different claims (especially the strong claims) regarding:

The Language-Specific Faculty: Triggering and Learning

A summary overview of background concepts relevant to critical period research will serve to set the stage for evaluating the findings. However, two distinct alternatives to the above proposed framework are logically possible:

These two alternative hypotheses should be kept in mind throughout the discussion if for no other reason than to help us evaluate the claims put forward, specifically by considering how they might be shown to be false.

To reiterate, it is the notion of a language-specific module or array of interacting language modules (related to alternative #2 above) that lies at the core of the model that we will examine. On the one hand, it is a common observation that young children construct the complex grammatical system of their primary language without any systematic training or explicit instruction. This kind of non-analytic and unconscious process seems to be confirmed by observed patterns of language use that vary widely from one culture to another, for example, in regard to the discourse style often described as "motherese," in which caretakers from certain socioeconomic groups and cultures attempt to modify language structures in child/adult interaction to increase comprehensibility and thus, hypothetically, foster language acquisition. On the other hand, "motherese" type modifications have not been observed in the socialization patterns of other cultures. Nevertheless, L1 acquisition proceeds at a remarkably uniform rate, regardless of these differences (Pinker, 1990).

More importantly, the highly complex and opaque nature of the rule systems that govern the child's developing grammar could not emerge from any kind of experience alone (systematic teaching or otherwise). The knowledge that underlies the structures of language that have developed by age six is simply too abstract to have been constructed from the application of general learning strategies, such as comparing and contrasting different kinds of examples (e.g., correct and incorrect), generalization, analogy, etc. The vast and elaborate knowledge of grammar that develops so uniformly and universally in preschool children requires no conscious reflection, explicit or indirect correction, or exposure to negative evidence. From positive evidence alone, which in addition is often incomplete, not relevant, or confusing (not all input provides correct examples of positive evidence), grammatical knowledge is built up. By six or seven years of age children command the basic structures of their primary language at a level that clearly and unambiguously differentiates them from all but the most successful second language learners of the language; this is what is referred to as "native-speaker grammatical competence." In the case of adults, for example, the application of general learning mechanisms such as inductive (problem solving) approaches do not guarantee uniform success in language learning; these general learning strategies applied by the cognitively immature child would be even less adequate for the task.

Thus, linguists propose that a language-specific faculty or module that is pre-experiential, innate, and genetically programmed places at the disposal of the child an elaborate conceptual framework of prior knowledge that assigns structure to linguistic input and builds the complex cognitive network which quickly develops into a complete knowledge system corresponding to the basic grammar of the child's mother tongue (Chomsky, 1988). In this sense, language proficiency is modular; some aspects such as the "core grammar" depend on the language-specific faculty and are universally acquired by all normal children, who in turn come to acquire them completely. Other aspects related, for example, to academic discourse and literacy may depend upon a more complex interaction among different cognitive domains outside of the language-specific faculty, including cognitive domains that correspond to general learning mechanisms. (See Maratsos [1992] on modularity in general, and Pinker [1996] on "modules of language" in relation to child language development.) Thus, "language proficiency" is a broader concept referring also to abilities that are learned, to using language for different purposes; some are "cognitively undemanding," as in skillful (e.g., socially appropriate), informal conversation in one's primary language, while others require the participation of more complex abilities and different kinds of knowledge structures. Precisely, it is this kind of mental heterogeneity, or compartmentalization, that can account for the specialized nature of first-language acquisition and can explain certain features of the acquisition process that are not shared by other aspects of cognitive development, universality and completeness being among the most noteworthy.¹ (See Lenneberg [1964] for an early discussion of criteria for innately given competencies that can be applied to language.)

Chomsky (1988) offers the analogy of a kind of circuit board in which the basic design is predetermined, but has the switches left "open" (representing the general principles of Universal Grammar, which for some authors is synonymous with Language Acquisition Device). This system of predetermined knowledge is available to all children prior to any contact with linguistic input. Thus, the circuit of "switches" requires limited positive evidence alone to be thrown toward one set of positions or another, i.e., toward one set of values of the parameters of UG or another:

From this perspective, then, to say that children "learn" their first language is misleading in some important respects. Rather, that part of grammatical knowledge that linguists term "core grammar" is triggered. In more technical terms, a series of specific parameter values are set. Thus, first language acquisition is, to a large extent, a process of choosing among options that have been genetically predetermined.²

Adults and Children: The Fundamental Difference Hypothesis

Our examination of the evidence will begin with a review of the research involving adult second-language learners. The availability of large numbers of international students studying in universities and colleges where English is the language of instruction has provided researchers with an important opportunity to examine the question of how second-language learning may be different from first-language acquisition.

More importantly, beginning with adults, the contrast with young children would be more evident. If the notion of a critical period is worth pursing at all, establishing a qualitative difference between the two most contrasting cases would be a good place to start. Presumably, if the capacity to acquire native-like grammatical competence begins to diminish at some point (necessarily, as we have shown in the previous section, after early childhood) adults would be good candidates for examining the extent of this diminution. In fact, the first hypotheses regarding a critical period estimated that puberty marked the point after which achieving complete native-speaker grammatical competence begins to elude the language learner (Lenneberg, 1967). On the other hand, if it could be demonstrated that the basic processes were the same in every way for the four and five year old acquiring his/her L1 and the 20-year-old ESL student, the inquiry would effectively end there.

Therefore, the first research question to be posed is, do adult second-language learners continue to have complete and unfettered access to Universal Grammar? Bley-Vroman (1989) reviews the evidence that suggests that adults can no longer avail themselves of the language-specific faculty, that L2 learning resembles general problem solving rather than the automatic type of parameter-setting process that is hypothesized in UG-driven L1 acquisition. In support of this distinction, the Fundamental Difference Hypothesis (FDH) calls attention to the following contrasts:

Bley-Vroman concludes that for the adult, general problem solving principles take the place of (or compensate for the erosion of) access to the language-specific module available to the young child, and that knowledge of the first language grammar fills the role which Universal Grammar plays in the acquisition of the primary language. Second language learners construct a kind of "UG-surrogate" which can only assure an approximation to native-level linguistic competence.³

Returning to the triggering-learning distinction, Schachter (1996) points out that the latter is characterized by procedures such as induction, through which knowledge is constructed more directly from the input; relevant prior knowledge supplies useful top-down organizing schemata, but the constraints (if they can be called that) are of a more general and open kind. With triggering, relevant prior knowledge allows the child to "select" from possible options the correct one. In this sense first language development proceeds by way of a kind of deduction. The triggering/deductive model assures completeness and universality—learning/inductive methods cannot assure either—although, depending on the circumstances, the L2 learner can approximate very closely, and perhaps even achieve, virtual completeness. Universality, however, would be excluded.

Delayed First-Language Learning

At this point, a review of the research conducted on late first language learning will offer another perspective on how certain aspects of language development are "biologically scheduled." The studies that are available are few given the highly abnormal circumstances in which these phenomena can be observed. One of the most reliable reports seems to be that of "Genie," the now well-known test case of first-language learning after puberty. As is typical of other such examples of pathologic L1 development, upon being liberated from total linguistic isolation at age 13 1/2, language learning proceeded at an accelerated rate, but development stabilized at a level far short of native-like competence. Suggestively similar to L2 learning, vocabulary knowledge showed greater development than morphology and syntax, with a wider gap separating receptive and expressive abilities as compared to native speakers (Long, 1990).⁴

While the evidence from case studies of modern feral children is entirely consistent with the notion of critical periods, drawing definitive conclusions is difficult. Aside from the limited number, it is impossible to separate out the general cognitive/emotional consequences of such extreme cases of child abuse. Keeping this admonition in mind, one may refer to the case reported in Hurford (1991) of a boy treated for severe combined immunodeficiency disease, after living in isolation from 9 months old to 4 years and 4 months old, who subsequently demonstrated near normal language development.

Age and Second-Language Learning

Considering what appears to be a fundamental difference between child L1 development and adult L2 learning, Lenneberg's demarcation of puberty as the maturational threshold seemed very plausible. However, a more complex research question is now posed: do children learning a second language benefit from free and unencumbered access to the Language Acquisition Device in a way that older, post-pubescent learners presumably cannot? The reader will take special note of how this question is different from the problem posed by Bley-Vroman, a distinction that is often left unspecified in the current discussion of critical periods. Here the key contrast is between two categories of L2 learners, not L1 versus L2.

At first glance the findings from research appear to be quite unfavorable to the idea of maturational constraints. Long's (1990) comprehensive review of the literature uncovers some contradictory evidence, but on the question of rate of L2 development, the majority of studies actually show an advantage for both older children and adults. That is, under controlled conditions, older students tend to attain mastery of one or another structural aspect of the target language more rapidly than younger students. However, according to Long, on the question of ultimate attainment the tables are turned; "younger starters" usually outperform older ones, given the opportunity (time) to "catch up." In studies of phonology for example, the beginning of a decline in the ability to master the sound system among immigrants is placed as early as age six.

In a study focused on syntactic knowledge in which age of arrival to the United States varied from childhood to adulthood, with minimum Length of Residence (LOR) of five years, the younger group, average Age of Onset (AO) of 8.6 years, performed better than immigrants arriving after 15, average AO of 27.1 (Patkowski, 1980). Significantly, younger subjects apparently revealed some non-native features during interviews in comparison to native-speaker (NS) controls who received "perfect scores" (an indication, parenthetically, that both interview items and NS control subjects were adequately selected). Long interprets the findings for rate and ultimate attainment as representing a short-term advantage for older learners. In reference to the puberty threshold, the results suggest rather a series of sensitive periods for second language—not just one—the first perhaps culminating around age six.

Bialystok and Hakuta (1994) urge caution regarding methodological considerations, specifically the interaction between LOR and AO and other intervening factors. Future research will need to control for actual length of time learning the L2 (an advantage conferred upon subjects with a early AO), cognitive maturity and test-wiseness of older subjects, and first-language background (typological closeness, level of prior higher order discourse abilities that tend to favor mastery of linguistic features, etc.).⁵

To sum up so far: the Fundamental Difference Hypothesis (Bley-Vroman, 1989; Schachter, 1996) distinguishes between normal L1 acquisition (that must coincide with early childhood) and adult L2 learning, focusing on two distinct processes at work—complete access to UG, and application of general learning strategies, respectively. Studies focused on L2 learners of different ages suggest some advantages for older learners (if the advantages do turn out to be temporary, the phenomenon still merits explanation) with long-term, ultimate attainment favored by age of initial contact with the L2: the earlier the contact the more complete the learning outcome in relation to target forms.

A Comparison of Native Competence, Delayed L1 Learning, and L2 Learning

We now turn our attention to a more detailed examination of a series of studies that are considered to be among the least ambiguous in regard to maturational factors in the domain of morphology and syntax. These studies include that undertaken by Newport and associates, and the extensive commentary of the findings by Bialystok and Hakuta (1994).

Two distinctly different populations were studied:

On the ASL Test Battery, Basic Word Order (subtest 1 out of 8) did not vary among the groups. Commenting on this result the author notes:

Indeed, on all the remaining seven tests of ASL morphology the results reflect significant differences in performance among all three groups. Especially noteworthy was the marked superiority of Native signers over Early learners. For Early learners, despite contact with ASL at 4-6 years of age, the delay was reflected in signing patterns that would be perceived as nonnative by a native signer, even after 30 years of everyday experience with ASL, a variable that was held constant for all subjects. Newport summarizes:

Also see Mayberry's (1993) study of native signers, late first language learners, and second language learners of sign, reporting results that generally confirm Newport's findings.

The second-language learners (the second population studied by Newport) were administered a grammaticality judgment assessment that tested for verb tense, pluralization, verb agreement (morphology), basic word order, permutations involving wh-questions, use of determiners and pronouns (syntax). See figures 1 and 2 in which the graphs for both the ASL group and the L2 group are redrawn. Again, in the case of the L2 learners we take note of the strong relationship between AO and test performance. Most significantly, however, a similar decline in ability begins to manifest itself among early, pre-pubescent learners; native speaker controls and the 3-7 AO group are indistinguishable one from the other, with the decline occurring between 3-7 and 8-10, not at puberty, and 11-15. As with the ASL native signers and learners, word order items did not discriminate between native-speakers and late learners.

Figure 1. Z-scores on tests of ASL morphology for native, early, and late learners of ASL

Redrawn from Newport (1990)

Bialystok and Hakuta (1994) regroup the age brackets and also argue that a sharp drop at puberty is not supported by the data, an interpretation, as we have seen, is consistent with other studies. Rather, comparing Figure 1 and Figure 2, the qualitative break, so to speak, appears to occur much earlier. In fact, the interpretation that seems to come forward for further reflection is that the 3-7 year "Age of Arrival" group may in effect be comprised of native speakers of English (i.e., English was their "second primary language," second merely chronologically, but primary in the more critical sense that access to UG was still fully available during the years when sufficient input was received in English, ages 3-7). The key demarcation turns out to be between native speakers and second language learners (including child L2 learners in the seven-to-puberty age range). Note that in the ASL study ability to attain complete native grammatical competence begins to diminish at the same time. Taking into account the results of studies that have suggested age effects in the opposite direction for L2 learning, Bialystok and Hakuta speculate that the concept of a critical period may not be applicable to second-language learning at all.⁷

Figure 2. Total score on test of English grammar in relation to Age of arrival

To recapitulate the arguments that have been considered up to this point: (a) a fundamental difference between child L1 and adult L2 establishes, in principle, the possibility of a critical period; (b) the original proposal that puberty marks a sharp decline in access to the LAD has been considerably weakened; (c) rather, the stage at which primary-language competence attains completeness around the basic grammatical structures (age 6, plus or minus 1) is suggested by more than one study as a possible threshold, such that the distinction being proposed now is, in effect, between first language (or first languages) and second language. Not coincidentally, this distinction concurs, hypothetically, with two periods: during the first, early childhood access to UG is complete. Henceforth, perhaps diminishing gradually, access may be less than complete.

In circumstances of anomalous language development, as in the case in which processible input in a natural human language is denied during early childhood when access to the LAD is normally unimpeded, language will develop without the full participation of the special language acquisition mechanisms of the LAD. This is the circumstance that corresponds to Newport's early (but not early enough) and late ASL learners.

Evidence from Canadian Immersion

Shifting our focus from experimental studies involving relatively small numbers of subjects to large-scale studies of second language learning, indirect evidence that is consistent with the early critical period hypothesis outlined above will be considered.

The Canadian French immersion programs have offered researchers the opportunity to examine the learning outcomes of large numbers of L2 students, under relatively uniform circumstances; the same L1, entering students are generally monolingual without significant preschool contact with French. In the case of the early total model, instructional programs provide significant amounts of input in the L2 for as long as eight years beginning at the same time (kindergarten). Of the different learning outcomes which investigators have studied, the ones that can potentially shed light on the question of maturational constraints are those that measured aspects of proficiency in French.

A review of the literature presents findings that have been remarkably uniform. While interpretations of the data have varied, sparking considerable debate, the facts are generally not in dispute:

The contrast between #2 and #3 is especially noteworthy. When it is possible to separate out aspects of language proficiency that depend on the application of general learning strategies (text-level organization, coherence, comprehension of subject-matter content, etc.) L2 learners, provided with adequate instruction, perform at the same level as native speakers. On measures that in fact distinguish between native speakers and L2 learners (pronunciation aside, primarily in the area of syntax and morphology), immersion students

reveal themselves unequivocally as the latter. While Hammerly's (1987) overall assessment of the programs is perhaps overly critical, his characterization of students' L2 grammatical knowledge is probably accurate—that of a stabilized interlanguage marked by fossilized patterns in relation to many structures that are acquired early and completely by native speakers.

It is important to emphasize that the evidence from early French immersion in favor of a critical period culminating around age six or seven is indirect and remains circumstantial. A number of possible explanations for the discrepancies come to mind; prominent in the literature, for example, is the lack of extra-curricular contact with the L2. Certainly, future research will need to untangle the various interacting factors. However, the apparent uniformity of French immersion students' level of interlanguage attainment is surely cause for reflection. For example, the quality and method of instruction would rank low on the scale of plausible explanations. If the principles and constraints of UG are still available (completely and in an unfettered manner) during the period from Kindergarten through eighth grade, 7,000 hours of communicative use and meaning-based exposure to the language should be sufficient for a significant minority of children to achieve, at least, near-native levels with "crisp and grammatically consistent forms," a result that continues to elude all variants of immersion. Returning to our model of how L1 core grammar is triggered, we can discard the hypothesis that preschool children must necessarily have the opportunity to systematically practice L1 language patterns, or engage in frequent and ongoing "negotiation of meaning" with more advanced speakers to be able to set the appropriate parameters. Indeed, some aspects of early language development (i.e., preschool acquisition of the mother tongue) probably do not require the "comprehensible" component of "comprehensible input," much less the communicative and interactive acquisition contexts of the "motherese" genre.

A possible UG-oriented explanation for the stabilization of French immersion interlanguage could hypothesize the following: Anglophone children enter Kindergarten during the final stages of a critical period that results in completion of the acquisition of L1 core grammar. The culmination of this process coincides with the beginning of a gradual diminution of access to the language-specific faculty. How rapidly and to what degree this erosion of access to UG unfolds remains unclear. Perhaps the rate and ultimate level of the erosion varies from child to child. Nevertheless, in general, children begin learning a second language (after exploiting the LAD for L1 development) with diminished resources available from the language module. At the same time, general cognitive development continues and the more mature strategies, now at children's disposal, can be applied to the task of L2 learning. The rate of intellectual development and the access to general, top-down, learning mechanisms also vary from child to child. The variation in available resources from the different modules and cognitive domains (some in decline, others in ascendancy) may account for individual differences in performance on linguistic tasks related to the target language grammar. Evidence in favor of this hypothesis would exist, for example, if immersion students' interlanguage attainment shows individual variation that is not typical of the more uniform development of grammatical structures among native speakers. In any case, the loss of complete access to UG, only partially compensated for by its "surrogate(s)," typically will place a limit on interlanguage development. We will recall, by the way, that the construction of an interlanguage is not the same as early childhood primary-language development (as well as the observation that all languages that develop sufficiently during early childhood are "first" or primary languages).

Credit goes to Selinker et al. (1974) for first describing immersion students' productive abilities in French in terms of the characteristics of interlanguage (IL): systematicity at any one particular stage, backsliding and reappearance of fossilized errors, stability over time of intermediate forms, and mutual intelligibility among speakers of the same IL. ⁸

A Proposal for a Comparative Study of Two Contrasting L2 Learner Populations

In North America and Mexico bilingual and second language teaching programs involving millions of school-age children have offered researchers important opportunities to focus attention on specific interacting factors precisely because the contexts of language learning differ, two cases in point: Modiano (1972) and Hernández-Chávez (1984). Since the differences are often broad and systematic, this allows for pertinent contrasts to be revealed for analysis and discussion. One such opportunity exists in a comparison between two large populations of L2 learners that share some key features in common, the French immersion students described above, and ESL students of the same age group (K-8) in the United States who receive systematic second language instruction of some kind: either or both (a) all English, ESL-style, immersion, and/or (b) dual-language instruction, but not (c) English mainstream submersion that provides minimal comprehensible input for the L2 learner.

Under the category of similar characteristics, the following would be relevant: (a) extensive contact with a second language in school, so that the L2 is both the object and medium of instruction; (b) in each case the L2 is a national and/or official language of the state or province and ( c) to keep L1 constant, the US cohort could be restricted to immigrants from Spanish-speaking countries (the selection of immigrants would offer the added advantage of limiting preschool exposure to the L2—in most cases prior contact with English would be minimal, thus approximating more closely the situation of French immersion starters (usually monolingual in their L1).

A number of differences are important to account for in the interpretation of any results:

Considering all the above, the following research question could provide a framework for designing a broad investigation of L2 learning outcomes focused on ultimate attainment of the respective linguistic systems, that is, how closely do children in the 6-to-puberty age group approximate native-speaker command of phonological patterns, morphology, syntax, and core lexical knowledge. The reader will note that the question of academic discourse proficiencies, especially literacy skills, would be considered separately, first, because a strict conceptual distinction must be maintained between the two aspects of general language ability (Francis, 2000). Measures of academic discourse proficiency do not and cannot discriminate between native-speaker competence and successively lower interlanguage levels of attainment. For example, it has been demonstrated that French immersion students' academic achievement in general and their command of discourse abilities in the L2 are comparable with both English-speaking peers (in regard to subject area knowledge) and French-speaking controls (in regard to both subject area content and discourse abilities). This finding, to reemphasize, is consistently and robustly demonstrated in all studies despite the equally consistent general failure to achieve even near-native levels in grammar. And secondly, it is the assessment of strictly linguistic structures that will reveal developmental patterns that are relevant to the issue of maturational constraints.

Turning now to the English L2 learners, a different profile, potentially, would emerge. Although a broad-based assessment of grammatical knowledge of ESL students of the type envisioned here has never been carried out, informal observation and numerous anecdotal accounts strongly suggest that, on average, immigrant children from Latin America who actually continue to reside continuously in the United States and are enrolled in school during the same period achieve significantly higher levels of L2 grammatical competence than their French L2 counterparts. This finding, to be confirmed empirically, would be entirely predictable; see "differences" under points #1 and #3 above.

The relevant questions include the following:

If the explanation for why French immersion students fall far short of native speaker target forms is that input is not sufficiently extensive or varied to allow the language faculty to construct a complete linguistic system, the litmus test, so to speak, for this claim would be ESL students with significant daily contact with the L2. In other words, if full access to UG is still available between six and puberty, it would be highly improbable that children with seven or eight years of daily interaction with English speakers and exposure to English-language television programming at such levels would not attain native-speaker levels. Thus, the standard would need to be very high for this group. Even if a significant minority was found to depart from native-speaker levels (and if this difference was manifested consistently even in what are often termed "surface features" of phonology, syntax and morphology, for example)⁹ the post-six-year-complete-access-to-UG claim would be seriously weakened.

The design of such a comparative study would need to be able to account for the relevant variables, and to the degree possible isolate the ones that are not pertinent to the research questions at hand. For example, the relative status of French and English in their respective countries (point #1 under "differences") cannot be controlled, but this factor favors English L2 learning, representing an intervening variable favoring complete attainment for the ESL group and skewing the tendencies in the same direction as difference #3 (the greater overall amount of L2 input that the English learners receive). In regard to difference #2, it has been generally recognized that French immersion students, being a self-selected population drawn from more favored social classes than the vast majority of Latin American immigrants, enjoy a wide range of advantages vis-à-vis school and schooling. Their impressive levels of achievement in subject areas and in English language arts are often attributed to this circumstance (Hernández-Chávez, 1984; Calvé, 1991; Valdés, 1997). Thus, to partially compensate for this imbalance, only those ESL students who exhibit consistent progress in subject area achievement would be selected for comparison. In this way, setting aside ESL students who fail to demonstrate progress in school would indirectly maximize "similar characteristic" #1 above: English learners with "extensive contact with a second language in school, L2 is both object and medium of instruction." Finally, in regard to this methodological issue, we should remember that the assessment criteria are to be restricted to linguistic or structural features of French and English, not academic discourse abilities. If access to UG is complete for the ESL student, even under conditions of average to marginally acceptable academic performance, extensive daily contact with the target language should be sufficient to ensure native levels of attainment in core grammar, even if report card grades are far from outstanding.

In summary, the following would represent evidence that access to the language-specific faculty, UG, begins to diminish (or be obstructed in some way) around age 6 or 7, and not at puberty:

In addition to the example of counterevidence mentioned above (the "qualitative break"), evidence favoring the hypothesis of "complete access to UG for L2 learning" during the 6-to-puberty period would include: a) beyond a certain threshold of extracurricular experience with English, all ESL students eventually attain native-speaker levels. If, on the other hand, beyond this hypothetical threshold only a fraction attains native speaker levels, the finding will continue to pose a problem for the hypothesis of "complete access to UG for L2 learning." Rather, a threshold would need to be identified beyond which complete acquisition of the target language is exhibited homogeneously; b) As with first-language development, if mastery of the basic grammatical patterns of English does not correlate with general academic achievement, this would suggest that grammatical competence developed thanks to the continued, undiminished functioning of the language-specific faculty (which operates largely independently of the general learning mechanisms). In other words, if ESL students' ultimate attainment were shown to be different in kind, and not just in degree, from French immersion students' attainment levels, this result would support the alternative of either a late-ending critical period (puberty), or none at all (continued access to UG into adulthood).

Conclusion

A number of bilingual pedagogy issues were mentioned in the introduction that would be directly or indirectly informed by further research on maturational constraints. Findings would also clarify key points in the current debate over school language policy.¹¹

1. When and in what proportion to introduce L2 instruction

None of the current hypotheses regarding critical periods suggests any negative consequences arising from the early introduction of second language instruction. Rather, a number of potential advantages favor it (see Krashen's (1996) "Gradual exit, variable threshold" model that recommends no postponement of English instruction). Also, Bialystok (1991) and Hakuta's (1990) claim that second-language learning promotes the development of metalinguistic awareness and higher order language proficiency would be compatible with an earlier critical period. Six and seven year olds who must begin to shift toward the application of general cognitive strategies for the task of L2 learning, because parameters are no longer automatically set by exposure to positive evidence only, will benefit from the new challenges that come with linguistic problem solving. Learning new grammatical structures will require the application of higher order cognitive operations, in this way promoting general intellectual development.¹²

On the other hand, assuming that age 6 or 7 represents a maturational milestone after which access to UG begins to diminish would also underline the rationale for preschool exposure to languages that are important for children to acquire and learn. For example, in bilingual communities whose local, ancestral language is the object of language revitalization efforts, unless children are in sustained contact with the local language from an early age (well before kindergarten), school-based programs, within the context of the predominant influence of the national, dominant language, will probably fall short of expectations. An earlier and complete access to the language faculty may also explain the phenomenon—reported in such situations of intense language contact—of a rapid development of L2 upon entry to primary school on the part of children who demonstrate productive, expressive abilities in only one language (Francis, 1997).

2. Rate at which bilingual students can be expected to develop control over L2 grammar

If the process of L2 learning after age 6 or 7 is different in some way from that of L1 acquisition, it would be a mistake to assume that it will proceed effortlessly, uniformly, automatically, and at the same rapid rate. Rather, teachers would expect a variation in both rate and ultimate attainment in accord with factors that are maturational, and in accord with factors related to the child's experience with and mastery of higher order learning strategies. If (school-age) child L2 learning is different in kind from early childhood L1 acquisition, this distinction offers yet another possible explanation for why English language learners may fail to make adequate progress in settings that do not provide significant amounts of high-quality ESL and/or dual-language instruction (e.g., varieties of "ESL-inclusion" that in effect resemble submersion, "ESL-pullout," etc.)

3. Rationale for L1 literacy development

As indicated earlier, higher order discourse abilities are largely independent of basic grammatical competence. However, initial literacy, for example, depends upon the learner's knowledge of the linguistic systems of the language through which reading and writing are to be taught. In the case of L2, a certain minimal threshold of control over grammatical patterns and vocabulary would be indispensable. For some children this threshold may turn out to be lower than for others. Nevertheless, the general principle would still apply to L2 learners as a whole. Thus if #2 is correct a major first language literacy component of bilingual, second language programs in the early years would ensure that all ESL students will have the opportunity to process texts with adequate grammatical knowledge at their disposal, not just those who have advanced most rapidly in English language development.

4. Communicative-based and form-focused teaching

Recent research, occasioned by the seemingly disappointing results in the area of French immersion students' L2 grammatical knowledge, has proposed a shift away from an exclusive emphasis on "comprehensive input" and toward models that provide for more explicit attention to language patterns, as in conscious reflection on structures in output, monitoring grammatical form, contrasting target forms with students' own partial knowledge, etc. (Swain & Lapkin, 1995; Schmidt, 1990). In contrast, see Krashen (1993) for a response to the "output hypothesis." If in fact conscious learning can play a significant role in child L2 development (and evidence suggests that it does), this poses an interesting question: if children after six and through puberty both profit from, and require the application of general cognitive strategies in L2 learning, this implies that a different kind of language development is involved—different from primary-language development that may "profit," in some way, from conscious learning, but clearly does not require it.

Most bilingual and ESL teachers today would favor the idea that conscious learning strategies and focus on form-learning activities can facilitate the English language development of their students. It turns out that this pedagogical intuition may be supported by sound theoretical principles.

References

Akmajian, A., Demmers, R., Farmer, A., & Harnish, R. (1995). Linguistics: An introduction to language and communication. Cambridge: MIT Press.

Bhatia, T., & Ritchie, W. (1999). The bilingual child: Some issues and perspectives. In W. Ritchie & T. K. Bhatia (Eds.), Handbook of child language acquisition (pp. 569-643). New York: Academic Press.

Bialystok, E. (1991). Metalinguistic dimensions of bilingual language proficiency. In E. Bialystok (Ed.), Language processing in bilingual children (pp. 113-140). New York: Cambridge University Press.

Bialystok, E., & Hakuta, K. (1994). In other words: The science and psychology of second language acquisition. New York: Basic Books.

Bley-Vroman, R. (1989). What is the logical problem of foreign language learning? In S. M. Gass & J. Schachter (Eds.), Linguistic perspectives on second language acquisition (pp. 41-68). New York: Cambridge University Press.

Bloom, P. (1994). Recent controversies in the study of language acquisition. In M. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 741-779). New York: Academic Press.

Calvé, P. (1991). Vingt-cinq ans d'immersion au Canada. Etudes de Linguistique Appliquée, 82, 7-23.

Chomsky, N. (1988). Language and problems of knowledge: The Managua lectures. Cambridge: MIT Press.

Cook, V. (1996). Competence and multi-competence. In G. Brown, K. Malmljaer, & J. Williams (Eds.), Performance and competence in second language acquisition (pp. 54-69). New York: Cambridge University Press.

Crawford, J. (1997). California's Proposition 227: A post mortem. Bilingual Research Journal, 21, 1-30.

Cummins, J. (1996). Negotiating identities: Education for empowerment in a diverse society. Ontario, CA: California Association for Bilingual Education.

De Gaetano, Y., Williams, L., & Volk, D. (1998). Kaleidoscope: A multicultural approach for the primary school classroom. Columbus, OH: Merrill.

Durkin, D. (1995). Language issues: Readings for teachers. White Plains, NY: Longman.

Faltis, C. (2001). Joinfostering: Adapting teaching for the multilingual classroom. Columbus, OH: Merrill.

Francis, N. (1997). Malinztin: Bilingüismo y alfabetización en la Sierra de Tlaxcala. Quito: Ediciones Abya-yala.

Francis, N. (2000). The shared conceptual system and language processing in bilingual children: Findings from literacy assessment in Spanish and Náhuatl. Applied Linguistics, 21, 170-204.

Genesee, F. (1991). L'immersion et l'apprenant défavorisé. Etudes de Linguistique Appliquée, 82, 77-93.

Genesee, F. (1994). Educating second language children: The whole child, the whole curriculum, the whole community. New York: Cambridge University Press.

Goldin-Meadow, S., & Mylander, C. (1990). Beyond the input given: The child's role in the acquisition of language. Language, 66, 323-355.

Gregg, K. (1996). The logical and developmental problems of second language acquisition. In W. Ritchie & T. Bhatia (Eds.), Handbook of second language acquisition (pp. 49-81). New York: Academic Press.

Hakuta, K. (1990). Language and cognition in bilingual children. In A. M. Padilla, H. H. Fairchild, & C. M. Valadez (Eds.), Bilingual education: Issues and strategies (pp. 47-59). Newbury Park, CA: Corwin Press.

Hammerly, H. (1987). The immersion approach: Litmus test of second language acquisition through classroom communication. The Modern Language Journal, 71, 395-401.

Harley, B., Allen, P., Cummins, J., & Swain, M. (1990). The development of second language proficiency. New York: Cambridge University Press.

Hernández, H. (1997). Teaching in multilingual classrooms: A teacher's guide to context, process, and content. Columbus, OH: Merrill.

Hernández-Chávez, E. (1984). The inadequacy of English immersion education as an educational approach for language minority students in the United States. In D. Dolson (Ed.), Studies on immersion education: A collection for United States educators (pp. 144-183). Sacramento, CA: California State Department of Education.

Hurford, J. (1991). The evolution of the critical period for language acquisition. Cognition, 40, 159-201.

Kowal, M., & Swain, M. (1997). From semantic to syntactic processing: How can we promote it in the immersion classroom? In R. K. Johnson & M. Swain (Eds.), Immersion education: International perspectives. (pp. 284-309). New York: Cambridge University Press.

Krashen, S. (1993). The case for free voluntary reading. The Canadian Modern Language Review, 50, 72-81.

Krashen, S. (1996). A gradual exit, variable threshold model for limited English proficient children. NABE News, 19, 1-18.

Krashen, S. (1999). Condemned without a trial: Bogus arguments against bilingual education. Portsmouth, NH: Heinemann.

Lapkin, S., Hart, D., & Swain, M. (1991). Early and middle French immersion programs: French language outcomes. The Canadian Modern Language Review, 48, 11-37.

LeBlanc, R. (1992). Les programmes d'immersion et l'habileté à communiquer. Etudes de Linguistique Appliquée, 88, 67-82.

Lenneberg, E. (1964). The capacity for language acquisition. In J. Fodor & J. Katz (Eds.), The structure of language: Readings in the philosophy of language (pp. 579-603). Englewood Cliffs, NJ: Prentice Hall.

Lenneberg, E. (1967). The biological foundations of language. New York: Wiley.

Lessow-Hurley, J. (2000). The foundations of dual language instruction. New York: Longman.

Long, M. (1990). Maturational constraints on language development. Studies in Second Language Acquisition, 12, 251-285.

MacSwan, J. (1999). A minimalist approach to intrasentential codeswitching. New York: Garland Publishing.

Maratsos, M. (1992). Constraints, modules, and domain specificity: An introduction. In M. Gunnar & M. Maratsos (Eds.), Modularity and constraints in language and cognition (pp. 1-23). Hillsdale, NJ: Lawrence Erlbaum.

Mayberry, P. (1993). First language acquisition after childhood differs from second language acquisition: The case of American Sign Language. Journal of Speech and Hearing Research, 36, 1258-1270.

Meisel, J. (1991) Principles of Universal Grammar and strategies of language use: On some similarities and differences between first and second language acquisition. In L. Eubank (Ed.), Point counterpoint: Universal Grammar in the second language (pp. 231-276). Philadelphia: John Benjamins.

Miramontes, O. (1993). ESL policies and school restructuring: Risks and opportunities for language minority students. The Journal of Educational Issues of Language Minority Students, 12, 77-95.

Modiano, N. (1972). Indian education in the Chiapas highlands. New York: Holt, Rinehart, and Winston.

Newport, E. (1990). Maturational constraints on language learning. Cognition, 14, 11-29.

Patkowski, M. (1980). The sensitive period for the acquisition of syntax in a second language. Language Learning, 30, 449-472.

Pawley, C. (1985). How bilingual are French immersion students? The Canadian Modern Language Review, 41, 865-876.

Pérez, B., & Torres-Guzmán, M. (1996). Learning in two worlds: An integrated Spanish/English biliteracy approach. White Plains, NY: Longman.

Pinker, S. (1990). Language acquisition. In D. Osherson & H. Lasnik (Eds.) Language: An invitation to cognitive science (pp. 199-241). Cambridge: MIT Press.

Pinker, S. (1994). The language instinct: How the mind creates language. New York: Harper Collins.

Pinker, S. (1996). Rules of language. In H. Geirsson & M. Losonsky (Eds.), Readings in language and mind (pp. 558-569). Oxford: Blackwell.

Piper, T. (1998). Language and learning: The home and school years. Columbus, OH: Merrill.

Ramírez, J. D. (1992). Executive summary: Longitudinal study of structured English Immersion Strategy, early-exit and late-exit transitional bilingual education programs of language-minority children. Bilingual Research Journal, 16, 1-61.

Roberts, I. (1994). Universal grammar and L1 acquisition. In N. C. Ellis (Ed.), Implicit and explicit learning of languages (pp. 455-476). New York: Academic Press.

Sacks, O. (1990). Seeing voices: A journey into the world of the deaf. New York: Basic Books.

Schachter, J. (1988). Second language acquisition and its relationship to universal grammar. Applied Linguistics, 9, 219-235.

Schachter, J. (1996). Learning and triggering in adult L2 acquisition. In G. Brown, K. Malmkjaer, & J. Williams (Eds.), Performance and competence in second language acquisition (pp. 70-88). New York: Cambridge University Press.

Schmidt, R. (1990). The role of consciousness in second language learning. Applied Linguistics, 11, 129-158.

Scovel, T. (1998). Psycholinguistics. Oxford: Oxford University Press.

Selinker, L., Swain, M., & Dumas, G. (1974). The interlanguage hypothesis extended to children. Language Learning, 25, 139-151.

Sharwood-Smith, M. (1994). Second language learning: Theoretical foundations. New York: Longman.

Singleton, D. (1995). A critical look at the critical period hypothesis in second language acquisition research. In D. Singleton & Z. Lengyel (Eds.), The age factor in second language acquisition (pp. 1-29). Clevedon, England: Multilingual Matters.

Spilka, I. (1976). Assessment of second language performance in immersion programs. The Canadian Modern Language Review, 32, 543-561.

Swain, M. (1984). A review of immersion education in Canada: Research and evaluation studies. In D. Dolson (Ed.), Studies on immersion education: A collection for United States educators (pp. 87-112). Sacramento: California State Department of Education.

Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive processes they generate: A step towards second language learning. Applied Linguistics, 16, 371-391.

Tager-Flusberg, H., & Calkins, S. (1990). Does imitation facilitate the acquisition of grammar? Evidence from a study of autistic, Down's Syndrome and normal children. Journal of Child Language, 17, 591-605.

Valdés, G. (1997). Dual-language immersion programs: A cautionary note concerning the education of language-minority students. Harvard Educational Review, 67, 391-429.

White, L. (1991). Second language competence versus second language performance: UG or processing strategies. In L. Eubank (Ed.), Point counterpoint: Universal Grammar in the second language (pp. 167-189). Philadelphia: John Benjamins.

Wong Fillmore, L. (1995). When learning a second language means losing the first. In G. González & L. Maez (Eds.), Compendium of research on bilingual education (pp. 19-36). Washington, DC: National Clearinghouse for Bilingual Education.

Endnotes

¹ See Lenneberg (1964) for an early discussion of criteria for innately given competencies that can be applied to language.

² For a more complete discussion see Akmajian et al. (1995), Scovel (1998) from a psycholinguistic point of view, Roberts (1994) on how linguistic input plus general learning strategies are insufficient, Goldin-Meadow and Mylander (1990), and Tager-Flusberg and Calkins (1990) on the concept of "creative construction" (in what way children's language development is "autonomous" from the input they receive), Schachter (1996) on setting parameters (triggering) versus learning, and Bloom (1994) for a general survey of the research on first-language acquisition.

³ For an alternative perspective that questions the Fundamental Difference Hypothesis, see Cook (1996), and White (1991), and for a comprehensive review of the literature, Sharwood-Smith (1994) and Bhatia and Ritchie (1999). Gregg (1996) discusses what has been termed the "logical problem" in L1 acquisition (also known as the problem of the "poverty of stimulus"): how to account for the "enormous gap between the grammatical knowledge acquired on the one hand, and the specific linguistic data the child receives on the other, such that the latter grossly underdetermines the former" (p. 52). In the case of L2 learning, Gregg suggests that the "logical problem" may still apply even if the majority of learners fail to achieve native-like competence. The final state of the interlanguage, however nontarget-like it turns out to be, may still require the participation of UG because L2 input data is simply not sufficient for the construction of such a complex system. The inquiry, then, would center on how to account for the vast differences between L1 acquisition (children "reach essentially the same end point from the same starting point,") (p. 56) and the wide variation in ultimate attainment in L2 learning. Related to this point, conclusive supporting evidence for the FDH does not hinge on generalized failure to achieve native-like competence since other learning mechanisms (outside of UG) may come to play a prominent role, in addition to linguistic knowledge represented in the learner's first language; his or her "starting point" would be a "UG-governed L1 grammar." In other words, the FDH does not imply that native-like performance in L2 is impossible to attain without full access to UG. Also see a discussion of the Fundamental Difference Hypothesis in MacSwan (1999, pp. 29-36), research evidence in Meisel (1991), and an application of the concept of modularity to L2 learning in Gregg (1996).

⁴ Also see Pinker (1994), and Sacks (1990) for a consideration of other cases within the framework of a broader discussion of the linguistic isolation most deaf children experience during early childhood.

⁵ In this paper "linguistics features" and "structural aspects" refer to the grammatical patterns of the language, corresponding to the rule systems that govern phonology, syntax and morphology, and semantic knowledge. Native speakers are considered to possess a "complete" control over these interacting systems in the sense that they do not exhibit partial knowledge that is typically revealed in the performance of most L2 learners. Complete knowledge that corresponds to this domain of language proficiency should not be confused with literacy-related academic uses of language that are learned in school or in other formal and informal learning situations.

⁶ See Sacks (1990) for why exposure to spoken language during early childhood for the "early" (but not "native") and "late" learners of ASL would not ensure normal acquisition of a first language.

⁷ See Singleton (1995) for a more extensive discussion from a different perspective (sharply critical of Bley-Vroman's FDH).

⁸ Also see Genesee (1991) for a discussion of the role of general cognitive abilities and the language faculty in different aspects of immersion students' L2 proficiency.

⁹ An important consideration in evaluating subjects for ultimate attainment of strictly linguistic features is that the domain of knowledge to be sampled (in spontaneous speech, grammatical acceptability judgement, etc.) must correspond to the subsystems of phonology, syntax and morphology, core lexical knowledge, etc., and not to the "higher order," literacy-related, discourse abilities. Consequently, the focus of assessment would be shifted to aspects of language that, from a teaching point of view, may be considered of marginal importance, the kind of discrete point errors (often termed "local errors") which often do not merit serious attention. Correctly so, pedagogically oriented error correction, for example, focuses on structures that are tied to the academic discourse objectives of the ESL curriculum. However, it is precisely the discrete point features that distinguish native-speaker levels from interlanguage patterns that clearly reflect partial grammatical knowledge.

Special attention would be given to the selection of judges and evaluators. Only bilinguals who are familiar with the variant of English spoken in the United States by Spanish speakers, who also often possess native-speaker competence in English, are able to discriminate between: (a) this particular variant of English, in which even Hispanic monolingual speakers of English maintain certain structural features transferred from Spanish (in speech), for example, to mark ethnic identity; and (b) true learner-language or interlanguage forms of English that reflect partial knowledge, unlike in #1.

¹⁰ In comparing measures of general academic achievement and progress in learning English, the relevant comparison (for the population proposed in this study and for the purposes that are of interest to us) would be between: (a) academic achievement in subject areas for which L1 (in this case Spanish) medium instruction was available, utilizing a Spanish language assessment instrument; and (b) command of the linguistic structures of English. Clearly, for the purpose of examining such a correlation, only ESL students who have had the benefit of bilingual instruction could be included in the study.

¹¹ See the discussion on the relative merits of different program options, the necessary features of L2 teaching that can ensure additive bilingual development, and the broader controversy on the role of primary-language instruction for English language learners in the United States (Ramírez, 1992; Miramontes, 1993; Valdés, 1997; Krashen, 1999). Clarity on the precise nature of the differences between L1 acquisition and L2 learning will also sharpen the arguments in favor of bilingual education at a time when confusion has deepened on even the more fundamental concepts (see Crawford, 1997).

¹² For an alternative perspective on childhood bilingualism, the reader is referred to Wong-Fillmore (1995) regarding the potential negative consequences of an early introduction of L2-medium content instruction.

Acknowledgements

The author extends thanks to the US/Mexico Fund for Culture and the Office of Grants and Contracts of Northern Arizona University for their generous support to the present research project.