Chapter01-11.pdf

The Critical Period Hypothesis and Second Language Acquisition

1. The Critical Period Hypothesis and Second Language Acquisition

Introduction

MuchresearchonL2acquisitioniseitherimplicitlyorexplicitlyinformedbytheCritical

Period Hypothesis, that is, the idea that maturational changes enforce an asymmetry

between language acquisition early in life compared to later acquisition. Originally

formulatedbyPenfield&Robert(1959)andLenneberg(1967),(versionsof)theCritical

Period Hypothesis have widely been recruited to explain both the success of child L1

acquisition and the comparative failure of adult L2 acquisition. This chapter introduces

theconceptofcriticalperiodsinbehaviouralandcognitivedevelopmentanddiscussesits

application to language acquisition. The major part of this chapter deals with critical

periodsinthecontextofL2acquisition.Afterdiscussingthecharacteristicsofpotential

critical periods and advancing a working definition of a critical period (Section 1.3), I

reviewtheempiricalevidenceadducedinfavourofandagainsttheexistenceofbounded

agerelatedconstraintsonL2acquisition(Sections1.3.2to1.3.5).

This chapter aims to give a general overview of the various ways in which the

investigation of agerelated effects in L2 acquisition has been tackled. Some studies

representative of each way will be discussed and their respective merits and lacunae

outlined. On the basis of this review, I introduce the conceptual basics and rationale of

generative approaches to the study of L2 ultimate attainment. These approaches will be

discussedwithaparticularemphasisontheacquisitionofphenomenaattheinterfacesof

syntaxwithothergrammaticalmodules.Severalrepresentativestudieswillbediscussed

toillustratetheoreticalapproachestolimitationsonadultL2acquisitionthatwillinform

theempiricalstudyofthisthesis.

1.1. Critical Periods: Features and biological foundations

A critical period describes a maturationally constrained period of time in which the

interactionofextantneurophysiologicalarchitectureandrelevantexternalstimuliresults

inthefulldevelopmentofaparticularcapacity. 1 Hence,acriticalperiodrepresentsatime

of heightened sensitivity to environmental stimuli for acquiring essential aptitudes that

willnotdevelopatallornotfullydevelopifexposuretotherelevantinputoccursearlier

orlaterthaninthebiologicallydeterminedtimewindow.

1 In the literature on maturationally governed developmental processes, the terms sensitive or optimal

period are often used (for discussion, see Ruben, 1997). A sensitive period denotes a less categorically

circumscribedperiodofheightenedsensitivitythansuggestedbytheclassicconceptionofacriticalperiod.

Inthefollowing,Iwillretainuseofthetermcriticalperiodforexpositoryconvenience(seealsoEubank&

Gregg, 1999; Hyltenstam & Abrahamsson, 2003), acknowledging, however, that the time span and the

slopeoftheonsetsandoffsetsofcriticalperiods(seeFigure1.1)mightbevariableandsmearoutacross

phenomena.

Chapter 1

Generically, then, a critical period is preceded by a period of low sensitivity to a given

stimuli and the subsequent critical period encompasses an onset, a peak period and an

offset,asillustratedinFigure1.1.Itissucceededbythesubsequentflatteningoutofthe

ability to acquire a behavioural process in response to the particular stimuli. Typically,

theonsetofacriticalperiodisratherabrupt,whiletheoffsetismoregradual;however,

different geometries of critical periods can be envisaged and have been put forth (see,

e.g.,Bornstein,1989,andBirdsong,2005afordiscussion).

PEAK

Figure 1.1.Geometry of prototypical critical period (adapted from Birdsong, 2005a).

Critical periods have been found to underlie the development of diverse neurosensory

capacities in biology (Knudsen, 2004). Critical periods conforming to the geometry in

Figure1.1havebeenobserved,forinstance,inthedevelopmentofvisioninferals(e.g.

Hubel & Wiesel, 1970), haptic perception in mice (Glazewski, Chen, Silva, & Fox,

1996),thematingsonginaspeciesofAfricanfrogs(Watson&Kelley,1992),auditory

orientation in the barn owl (Knudsen & Knudsen, 1990) and the song of various bird

species,e.g.zebrafinchesandsparrows(Marler,1990).

Perhaps the closest analogue to the development of human language, the

development of speciesspecific birdsong, has been demonstrated to be contingent on

exposure to the song of conspecifics in the first year of life of certain bird species.

Further,youngbirdsmustbeabletohearthemselvesduringthattimeinordertodevelop

normalsongpatterns(e.g.Brenowitz,Margoliash&Nordeen,1997).Iftheseconditions

arenot met,thebirds willdevelopabnormal songpatterns that willnotbequalitatively

affected by later exposure to conspecific song patterns. Experimentally, it has further

beenshownthattheemergenceandthecharacteristicsofbirdsongcanbemodulatedby

direct neurophysiological intervention (e.g. lesioning brain areas dedicated to song

learning)orhormonalmanipulation,whichsuggeststhattheonsetandoffsetofacritical

periodforbirdsongisneurallyfoundedandhormonallyregulated(ibid).

CRITICALPERIOD

The Critical Period Hypothesis and Second Language Acquisition

In the human development of hearing, recent studies on auditory perception and

speechcomprehensionofcongenitallydeafchildrenwhohavebeenfittedwithcochlear

implantsatvariousagessuggestthatagemediatesthesuccessfulacquisitionofauditory

discrimination skills. Cochlear implants are surgically implanted devices that transform

auditory signals into electrical pulses to stimulate the auditory nerves in individuals

suffering from severe to profound hearing loss in order to emulate environmental

auditory stimuli, in particular speech. Several studies finda nonlinear relation between

age of implantation and the development of auditory skills. They suggest that normal

speech recognition development is contingent upon auditory input during a constrained

critical period in the first few years of life (e.g. McConkey Robbins, Burton Koch,

Osberger,ZimmermanPhillips&KishonRabin,2004;Sharma,Dorman&Spahr,2002;

for critical discussion, see Harrison, Gordon & Mount, 2005). In view of these critical

periodsintheacquisitionofsensorimotorcapacitiesservinglanguagecomprehension,it

is natural to ask whether critical periods can also be found for the neurocognitive

representationsoflanguage.

1.2. The Critical Period Hypothesis and language acquisition

OriginallyformulatedbyPenfield&Roberts(1959)andfurtherelaboratedbyLenneberg

(1967), the idea that a constrained period of heightened neural sensitivity underlies

language development has become widely espoused in research on first language

acquisition to account for the contrast in speed, ease and success of normal child

language development and the failure in later language acquisition when no relevant

linguisticinputwasavailableduringearlychildhood.

Evidence to underscore the reality of critical periods for language acquisition is

furnished by (a) feral children (for review, e.g. Curtiss, 1988), (b) congenitally deaf

children acquiring sign language later in life (e.g. Mayberry & Lock, 2003) and (c)

differences in auditory discrimination abilities of nonnative phonemic contrasts in

childrenatdifferentages(e.g.Werker&Tees,2005).

Although a critical period is often ascribed to language acquisition in general,

there has been growing awareness that ‘language’ is too coarsegrained a concept to

whichtoattributematurationaleffectsin toto(e.g.Eubank&Gregg,1999;Long,1990;

Newport et al., 2001). Seeing that ‘language’ consists of multiple conceptually and

neurallydistinguishablesubcomponentssuchassyntax,semantics,phonology,etc.,each

of these components could either be subject to critical periods of potentially different

properties,scopesandtemporalcharacteristicsorbesparedfrommaturationalconstraints

altogether.

Support for a differentiated conceptualization of multiple critical periods in

languageacquisitioncanindeedbefoundinmanyofthestudiesdocumentinglinguistic

failureindelayedfirstlanguageacquisition.Forinstance,thewellknowncaseofGenie,

Chapter 1

whowascompletelydeprivedoflinguisticinputfromtheageof1;6to13years(Curtiss,

1988), illustrates dissociations between linguistic components. After rescue, Genie

showedquickdevelopmentofvocabularyandcommunicativeskills,whereashersyntax

and intonation remained rudimentary even after prolonged exposure (for further

examples,seePinker,1994:290296;Singleton&Ryan,2004:3160).

More recent studies furnish evidence that even subcomponents of linguistic

domains, such as syntax or phonology, can be differentially affected by maturational

constraints.Inrecentworkonlateacquiredsignlanguage,Mayberry&Lock(2003;see

also Mayberry & Lock, 2002) report on a comparative study of congenitally deaf

individuals who had no exposure to (signed) linguistic input prior to age six and were

testedmorethan11yearsafterthedelayedonsetoftuitionandusageofEnglishasadults.

Their linguistic knowledge was compared to three groups of adults who had received

spoken or signed English input or spoken nonEnglish (Urdu) input in early childhood

and had had comparable lengths of exposure to English. On several tasks, the

congenitally deafindividualswith noearly exposure to languagecorrectly judged some

syntacticviolations(ungrammaticalauxiliaryselection,ungrammaticalobjectplacement,

ungrammatical conjunction) but performed at chance on identifying ungrammatical

structures involving syntactic movement (passives, relative clauses). By contrast, the

other groups showed no such asymmetry in behaviour. For phonology, findings of a

similar nature indicating different age effects depending on the type of phonological

phenomena have been reported in the context of cochlear implantation by Harrison,

Gordon&Mount(2005),andinthecontextofnormalchilddevelopmentbyWerker&

Tees(2005).

In sum, the robust findings that humans with no early exposure to language

demonstrate persistent failure in late language acquisition suggest that critical periods

constrain language development. However, different domains of language are

differentially affected by early deprivation of input, with some showing better

recoverabilityupondelayedexposurethanothers.

1.3. The Critical Period Hypothesis and second language acquisition

1.3.1. Conceptualizing the Critical Period Hypothesis in L2 acquisition

Conceptualizing a critical period in the context of adult second language acquisition

requiressevererevisionstothewaycriticalperiodsarecommonlycharacterized(Eubank

&Gregg,1999).Incontrasttotheprototypicalcasesofcriticalperiods,wherenorelevant

external stimuli were available at critical moments, adult L2 acquisition by definition

occurs against the background of successful L1 acquisition during the critical period.

Hence,anadultL2learnercannotbesaidtohavemissedthecriticalperiodforlanguage

acquisitionoutright;atthesametime,acquiringtheL2indeedoccursoutsidethecritical

The Critical Period Hypothesis and Second Language Acquisition

period. Assuming, then, that L1 acquisition entails a particular neurocognitive

organization of linguistic knowledge, acquiring an L2 in adulthood requires changes to

the extant patterns established for the L1. Irrespective of the conceptualization of these

changes, such neurocognitive reorganization implicates much more limited changes in

neural architecture and cognitive processes compared to the largescale organization of

linguistic knowledge that underlies the transition from the child’s prelinguistic state to

its mature state (e.g. Lenneberg, 1967: 176). As a consequence, e.g. Eubank & Gregg

(1999)cautionthatthebehaviouralreflexesofcriticalperiodeffectsobserveableinadult

secondlanguageacquisitionshouldberelativelyminorinscope.

Againstthisbackground,criticalperiodeffectsalsoneedtobedifferentiatedfrom

ageeffectsinL2acquisitionthatarenotmaturationallybounded.LateL2acquisitionis

likelytobesusceptibletogradualeffectsofcognitiveagingthatbegininearlyadulthood.

Cognitive aging affects language performance in causing slower and less efficient

informationprocessingandreducedworkingmemorycapacities(e.g.Park,2000).Inand

ofthemselves,cognitiveeffectsofagingmayimplicatelesssuccessfuloutcomesinadult

L2 acquisition (e.g. Bialystok & Hakuta, 1999; Hakuta, Bialystok, & Wiley, 2003).

Crucially, these age effects should be continuous and show a linear relation to age as

opposed to the nonlinear relation of age and acquisition predicted by a critical period

(Figure1.1).

In addition, the componential organization of language potentially entails that

theremightbedifferentcriticalperiodsfordifferentlinguisticsubcomponents,withsome

subcomponentsshowingnocriticalperiodsandothersbeingsubjecttocriticalperiodsof

different onsets and offsets. In recent reviews of the literature on critical periods in L2

acquisition,Long(2005)andSingleton(2005)giveoverviewsofthevaryingendsofthe

offset that have been proposed for subparts of language. For phonetics and phonology,

theproposalsrangefromoneyear(e.g.Ruben,1997)totwelveyearsofage(e.g.Scovel,

1988),formorphosyntax,thesuggestionsextendfromshortlyafterbirth(Hyltenstam&

Abrahamsson,2003)tofifteenyearsofage(Long,1990;2005).Table1.1(adaptedfrom

Singleton,2005)presentsaschematicoverviewofsomeproposals.

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: