Second Language Acquisition and the Critical Period Hypothesis

David Birdsong

• Publisher: Routledge
• Year: 1999
• ISBN:
  9781135674885
• Edition: 1
• Topic: Education
• Subtopic: Education Theory & Practice

Second Language Acquisition and the Critical Period Hypothesis is the only book on the market to provide a diverse collection of perspectives, from experienced researchers, on the role of the Critical Period Hypothesis in second language acquisition. It is widely believed that age effects in both first and second language acquisition are developmental in nature, with native levels of attainment in both to be though possible only if learning began before the closure of a "window of opportunity" – a critical or sensitive period. These seven chapters explore this idea at length, with each contribution acting as an authoritative look at various domains of inquiry in second language acquisition, including syntax, morphology, phonetics/phonology, Universal Grammar, and neurofunctional factors. By presenting readers with an evenly-balanced take on the topic with viewpoints both for and against the Critical Period Hypothesis, this book is the ideal guide to understanding this critical body of research in SLA, for students and researchers in Applied Linguistics and Second Language Acquisition.

CHAPTER ONE
Introduction: Whys and Why Nots of the Critical Period Hypothesis for Second Language Acquisition

David Birdsong

University of Texas

The facts of adult second language acquisition (L2A) contrast sharply with those of first language acquisition (L1A). Whereas the attainment of full linguistic competence is the birthright of all normal children, adults vary widely in their ultimate level of attainment, and linguistic competence comparable to that of natives is seldom attested. A reasonable explanation for the facts of L1A and L2A is given by the Critical Period Hypothesis (CPH). In its most succinct and theory-neutral formulation, the CPH states that there is a limited developmental period during which it is possible to acquire a language, be it L1 or L2, to normal, nativelike levels. Once this window of opportunity is passed, however, the ability to learn language declines. Consistent with the CPH are the morphological and syntactic deficits of Genie, who was largely deprived of linguistic input and interaction until age 13 (Curtiss, 1977), as well as the desultory linguistic achievements of most adult L2 learners.

With a focus primarily on L2A, the present volume explores reasons why humans might be subject to a critical period for language learning. It also examines the adequacy of the CPH as an explanatory construct, the “fit” of the hypothesis with the facts.

To both of these dimensions, the contributors offer cutting-edge thought and experimentation. In examining the possible causes of a critical period for L2A, the researchers bring the CPH into line with specifics of recent linguistic theory (Eubank & Gregg, chap. 4), discern neurofunctional differences between early- and late-learned language (Weber-Fox & Neville, chap. 2), and suggest sources of limits to language learning that are accommodated in modern evolutionary thinking (Hurford & Kirby, chap. 3). In questioning the explanatory suitability of the CPH-L2A, contributors bring new empirical data and argumentation to bear on matters once thought to be settled, such as the heuristic utility of the CPH-L2A (Flege, chap. 5), the shape of the age function, in theory and in fact (Bialystok & Hakuta, chap. 7), and the possibility of nativelike attainment in L2 pronunciation (Bongaerts, chap. 6).

These two approaches—one that ponders the etiology of a critical period for L2A and the other that disputes the adequacy of the CPHL2A— are representative of current intellectual discourse. In according equal time to each of the approaches, this volume aims at a balance of scholarship pro and contra the CPH in the L2A context.

As a prolegomenon to these chapters, it is instructive to examine a few of the more prevalent formulations of the CPH-L2A, looking in particular at the proposed mechanisms of age-related effects. The introduction will also situate this book within the current intellectual climate of questioning the received wisdom relating to the CPH-L2A.

THE WHYS: VIEWS ON THE ONTOGENY OF TIME-BOUNDED SUCCESS IN L2A

Earlier references to “the” CPH are somewhat misleading, for there is no single CPH.¹ Rather, there are varied formulations, each of which takes a different ontogenetic tack on the limits of language acquisition. It is customary, however, to refer to them collectively, because, manifestly, they share the common denominator of determinism. That is, they assume a nonnativelike end state for late language acquisition and seek explanations for this outcome in developmental factors that inevitably affect all members of the species.

In this volume, each of the chapters addresses at least one of several critical period hypotheses as they apply to adult L2A. As a preview of these varied formulations of the CPH—and as an introduction to other, kindred proposals not mentioned in the chapters—this section offers sketches of some of the mechanisms that researchers have proposed as underlying age-related declines in language learning ability.

Loss of Neural Plasticity in the Brain

Because of progressive lateralization of cerebral functions and ongoing myelination in Broca’s area and throughout the cortex, the neural substrate that is required for language learning is not fully available after the closure of the critical period. This formulation was originally proposed by Penfield and Roberts (1959), and later popularized by Lenneberg (1967), who postulated that the end of the critical period was marked by “termination of a state of organizational plasticity linked with lateralization of function” (p. 176). Variations on this line of thinking have been advanced for the L2A context (e.g. Long, 1990; Patkowski, 1980; Pulvermüller & Schumann, 1994; Scovel, 1988).

Lenneberg (1967) directed most of his argumentation to primary language acquisition. However, he made a brief foray into L2A and pointed to learners’ progress as well as their shortcomings. Here, Lenneberg moved from brain-based to mind-based commentary, alluding to an appendix in his book—written by Chomsky—that outlines Universal Grammar (UG)-based formal similarities among natural languages. For adults learning an L2, Lenneberg (1967) invoked the presence of this mental “matrix for language skills” to square the facts of (partial) L2A success with closure of the critical period:

Most individuals of average intelligence are able to learn a second language after the beginning of their second decade…. A person can learn to communicate in a foreign language at the age of forty. This does not trouble our basic hypothesis on age limitations because we may assume that the cerebral organization for language learning as such has taken place during childhood, and since natural languages tend to resemble one another in many fundamental aspects (see Appendix [A]), the matrix for language skills is present. (p. 176)

For related thinking about the linkage of neurological development and the mental representation of UG, see Eubank and Gregg (chap. 4, this volume) and Jacobs (1988).

Loss of (Access to) the Language Learning Faculty

The closure of the critical period entails a loss of UG, a mental faculty consisting of innately specified constraints on the possible forms that natural language grammars may take. A weaker version of this approach suggests that UG continues to be mentally represented but for various reasons is no longer available or accessible to the language learner. It should be noted that, because the L1 grammar is an instantiation of UG (see previous section), one can plausibly account for at least some of the headway that learners do make in L2A.

With the offset of the critical period, there may also be a loss of innate learning strategies presumed specific to the learning of language. These include the Subset Principle, which guides the learner to posit the most conservative grammar consistent with the linguistic input. By hypothesis, these epistemological components are the sine qua non of language acquisition; their absence essentially guarantees failure to attain nativelike competence. Thus the Fundamental Difference Hypothesis (Bley-Vroman, 1989) attributes the divergent end states of early L1A and late L2A to loss of, or lack of access to, UG and associated learning principles.

Principled inquiry concerning the role of UG in both the initial and end states of adult L2A comes in many forms (for a recent selection, see Flynn, Martohardjono, & O’Neil, 1997). One prominent line of thinking holds that invariant principles of UG are not lost in adult L2A; rather, what is problematic is the acquisition of L2 parameters: “Parameter values become progressively resistant to resetting with age, following the critical period” (Towell& Hawkins, 1994, p. 126). Simplistically, the difficulty in resetting parameters resides in having to “unlearn,” in the sense of relinquishing the representation of a parameter having a unique, L1-based setting, and establishing in its stead a biunique setting compatible with both the L1 and the L2 (for elaboration on parameter resetting, see Eubank and Gregg, chap. 4, this volume). In a later section, I summarize a contrasting approach to unlearning under the connectionist model of acquisition.

Maladaptive Gain of Processing Capacity with Maturation

As children develop, they are increasingly capable of processing linguistic input. However, Newport (1990, 1991) argued that cognitive immaturity, not cognitive maturity, is advantageous for language learning. Young children’s short-term memory capacity allows them initially to extract only a few morphemes from the linguistic input. Working within these processing limits, children are more successful than adults, whose greater available memory allows for extracting more of the input, but who then are “faced with a more difficult problem of analyzing everything at once” (Newport, 1991, p. 126). The benefits of starting small have been demonstrated in simulations of the acquisition of English morphology (Goldowsky & Newport, 1993). Similarly, Elman’s (1993) connectionist model starts with limited memory, then undergoes maturational changes (incremental increases in memory capacity). Training of networks under this condition succeeds in processing complex sentences. If the starting point is a fully formed adult-like memory, however, the complex sentences are not successfully processed by the network.

This “less is more” formulation of the CPH is apparently not confined to the domain of language acquisition: “The more limited abilities of children may provide an advantage for tasks (like language learning) which involve componential analysis” (Newport, 1990, p. 24; italics added). Nor is any loss of an innate language learning faculty implied: “the language acquisition capacity remains intact, but as children mature beyond the ages of four or five its function is impeded by the child’s increasingly sophisticated cognitive abilities” (Meier, 1995, p. 613). In a similar vein of thought that specifically targets L2A, Felix’s Competition Model (e.g., Felix, 1985) posits the coexistence of an intact UG and advanced domain-general cognition, and maintains that competition between the two systems results in victory for the latter. Mature domain-general cognition is thought to be ill-suited to the narrow, modularized task of acquiring language, hence the lack of success typically associated with adult L2A. The inappropriateness of certain mature cognitive mechanisms in the L2A context was explored by Birdsong (1994) and Bley-Vroman (1989).

Rosansky (1975) appealed to a Piagetian developmental model of cognition and argued that the emergence of Formal Operations during adolescence might forestall language learning. Although Rosansky’s theoretical constructs differ from those of Newport, the reasoning of the two researchers is remarkably similar. For Rosansky (1975),

initial language acquisition takes place when the child is highly centered [i.e., in stages prior to Formal Operations]. He is not only egocentric at this time, but when faced with a problem he can focus (and then only fleetingly) on one dimension at a time. This lack of flexibility and lack of decentration may well be a necessity for language acquisition. (p. 96)

Use It Then Lose It

After childhood, unneeded neural circuitry and the language learning faculty it underlies are “dismantled” because the relevant neural tissue incurs metabolic costs (Pinker, 1994). This reasoning, whereby early language learning is biologically favored over later learning, is rooted in modern evolutionary thinking. Early learning of language is preferred in order that we may reap the benefits of linguistic communication over a longer stretch of our lifetime. So whereas our use of language continues through adulthood, the language learning faculty has served its purpose early on. To retain it would be uneconomical.

The evolution of our species has taken account of this one-shot utility. As Pinker (1994) argued:

Language-acquisition circuitry is not needed once it has been used; it should be dismantled if keeping it around incurs any costs. And it probably does incur costs. Metabolically, the brain is a pig. It consumes a fifth of the body’s oxygen and similarly large portions of its calories and phospholipids. Greedy neural tissue lying around beyond its point of usefulness is a good candidate for the recycling bin. (pp. 294–295)

Hurford (1991) similarly accommodated the “use it then lose it” version of the language learning faculty within an evolutionary model: “The end of the critical period at around puberty is…a point where the selection pressure in favour of facilitating factors ceases to operate, because of success at earlier lifestages…. The ‘light’ goes out for lack of pressure to keep it ‘on’” (p. 193).

Pinker (1994) speculated that the critical period for language acquisition is evolutionarily rooted in the more general phenomenon of senescence. Natural selection asymmetrically favors young organisms over older ones, assigning to youth the emergence of the lion’s share of genetic features, which deteriorate at differing rates with increasing age. Using the example of lightning striking and killing a 40-year-old, Pinker noted that if a bodily feature had been designed to emerge after the age of 40, it would have gone to waste:

Genes that strengthen young organisms at the expense of old organisms have the odds in their favor and will tend to accumulate over evolutionary timespans, whatever the bodily system, and the result is overall senescence. Thus language acquisition might be like other biological functions. The linguistic clumsiness of tourists and students might be the price we pay for the linguistic genius we displayed as babies, just as the decrepitude of age is the price we pay for the vigor of youth. (p. 296)

Use It or Lose It

On the mental muscle metaphor, the language learning faculty atrophies with lack of use over time. Paltry progress in postadolescent L2A is clearly compatible with this view. Further, deriving from “use or lose” the inference that if the language learning faculty is used it will not be lost, this “exercise hypothesis” can also accommodate anecdotal accounts of individuals who start L2 acquisition early and continue to acquire foreign languages successfully into adulthood.

The exercise hypothesis was elaborated in greatest detail by Bever (1981). Under Bever’s view, for acquisition of a given linguistic structure to take place, the systems of speech production and speech perception should work in tandem. In the absence of ongoing language learning activity, however, the two systems become progressively independent (with perceptive abilities outstripping productive abilities), because the psychogrammar, which normally mediates production and reception, ceases to function. (Bever’s psychogrammar may be likened to a combination of UG, plus an organizer of acquired linguistic knowledge, plus an equilibrator of production and reception capacities at the moment of acquisition of a given structure.) Under conditions of continual use, however, the psychogrammar does not cease to function, and production and perception do not dissociate:

So long as one is continually learning a new language the systems of production and perception never become fully autonomous, and closed off from each other. That is, continuous acquisition can stave off the independence of the systems, and therefore delay the apparent critical period. (Bever, 1981, p. 194)

Whereas the use it or lose it formulation predicts that critical period effects can be skirted under conditions of continued language learning, the “use it then lose it” version would seem to imply inevitable loss of language learning ability at the offset of the critical period. The two conceptions also differ in terms of the postmaturational fate of language learning circuitry. For Pinker, natural selection eliminates the metabolically hungry but functionally obsolete language learning mechanism. For Bever (1981), the psychogrammar “does not disappear after its usefulness is past because it is so entrenched as a mental system”; rather, it hangs around, taking an enormous metabolic toll: “The psychogrammar is not a joy of adulthood, but a burden, an adventitious relic left over from a dozen years of language learning” (p. 188).

Learning Inhibits Learning

In connectionist networks, learning is a matter of progressively accumulating and strengthening input-output associations. The strength of an association is functionally a probabilistic weighting corresponding to the likelihood that a given output of the system is correct. One downside to this kind of learning is that it is difficult to undo: As Elman et al. (1996) noted, “across the course of learning…the weights within a network become committed to a particular configuration…. After this ‘point of no return’ the network can no longer return to its original state” (p. 389).

Consider the example of the word-final phonemic sequence /oral vowel +n/ in French, which is strongly correlated with feminine gender in nouns and adjectives. Under the connectionist model, an adult native French speaker develops a high weighting for the cooccurrence of this sequence with feminine gender. Once the weighting has become stable, it is difficult to perform the unlearning required for representational reorganization. So, if the French native encountered word-final /oral vowel +n/ in a foreign language, the learner’s initial assumption would be that the gender of the word is feminine. This functional state would persist despite input to the system about the inadequacy of its o...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Credits
• Preface
• Chapter One: Introduction: Whys and Why Nots of the Critical Period Hypothesis for Second Language Acquisition
• Chapter Two: Functional Neural Subsystems Are Differentially Affected by Delays in Second Language Immersion: ERP and Behavioral Evidence in Bilinguals
• Chapter Three: Co-Evolution of Language Size and the Critical Period
• Chapter Four: Critical Periods and (Second) Language Acquisition: Divide et Impera
• Chapter Five: Age of Learning and Second Language Speech
• Chapter Six: Ultimate Attainment in L2 Pronunciation: The Case of Very Advanced Late L2 Learners
• Chapter Seven: Confounded Age: Linguistic and Cognitive Factors in Age Differences for Second Language Acquisition