Download - Recent developments in SLA studies of phonology. Evaluating the “Critical Period” Hypothesis: Perceptual Learning of Mandarin Tones in American Adults.

Transcript

Recent developments in SLA Recent developments in SLA studies of phonologystudies of phonology

Evaluating the “Critical Period” Hypothesis:Evaluating the “Critical Period” Hypothesis:PerceptualPerceptual L Learning of Mandarin Tones in earning of Mandarin Tones in American Adults and AmericanAmerican Adults and American C Children at 6,hildren at 6, 10 and 14 Years of Age.10 and 14 Years of Age.

Yue Wang and Patricia K. KuhlYue Wang and Patricia K. Kuhl

abstractabstract

examinexamination ofation of the perceptual learning of the perceptual learning of lexical tones in Mandarin Chinese by lexical tones in Mandarin Chinese by

American youngAmerican young adultsadults andand

children from 6 to 14 years oldchildren from 6 to 14 years old

covering the agecovering the age range surrounding the range surrounding the “critical period”“critical period”

tthe participantshe participants received a two-week received a two-week computerized Mandarin tone trainingcomputerized Mandarin tone training program designed to be child-friendlyprogram designed to be child-friendly

for the trainees in each of the four age for the trainees in each of the four age groups groups ((but notbut not the controlsthe controls)), percent , percent correct identification correct identification increasedincreased significantlysignificantly from the pre-training test to the post-from the pre-training test to the post-trainingtraining testtest

comparing the pre-pubertycomparing the pre-puberty and the post-and the post-puberty groups, puberty groups, nono abrupt abrupt decrease in the decrease in the degree of improvementdegree of improvement was found was found, as , as would have beenwould have been predicted by the Critical predicted by the Critical Period HypothesisPeriod Hypothesis

tthese resultshese results support the view that support the view that language learning is not a strictlylanguage learning is not a strictly timed timed developmental process with rigid cut-off developmental process with rigid cut-off periodsperiods

predictionspredictions

aaccording to the Critical Period Hypothesis ccording to the Critical Period Hypothesis (CPH),(CPH), language-learning ability is language-learning ability is reduced at pubertyreduced at puberty

resultingresulting from the loss of neurological from the loss of neurological plasticity of the brain plasticity of the brain (Lenneberg 1967)(Lenneberg 1967)

puberty marks a closing of the window ofpuberty marks a closing of the window of opportunity for learning opportunity for learning

aan alternative theoryn alternative theory::

futurefuture learning may be limited by prior learning may be limited by prior learninglearning

which itselfwhich itself alters the brainalters the brain

producing a kind of interference effect thatproducing a kind of interference effect that impacts later learning impacts later learning (Kuhl 2000)(Kuhl 2000)

thus, thus, experienceexperience itself, rather itself, rather than time, is than time, is the critical variable the critical variable

recent studies have indicated that corticalrecent studies have indicated that cortical representations may be continuously shaped representations may be continuously shaped throughout lifethroughout life ( (Van Turennout Van Turennout et al. 2000)et al. 2000)and that language-related cortical responses and that language-related cortical responses may differmay differ as a function of linguistic experience as a function of linguistic experience ((PeraniPerani et al.. 1998) et al.. 1998) with sufficient experiencewith sufficient experience and exposure, adult and exposure, adult second language (L2) learners cansecond language (L2) learners can authentically authentically perceive or produce novel L2 sounds whichperceive or produce novel L2 sounds which have no native language (L1) phonemic have no native language (L1) phonemic counterparts counterparts ((FlegeFlege 1995) 1995)mmore explicitly, auditory training studies have ore explicitly, auditory training studies have successfullysuccessfully trained adult listeners to better trained adult listeners to better distinguish non-nativedistinguish non-native sounds sounds ((LoganLogan et al. et al. 1991)1991)tthese findings suggest that adult perceptualhese findings suggest that adult perceptual mechanisms have more plasticity than was mechanisms have more plasticity than was previouslypreviously recognizedrecognized

few direct testsfew direct tests of of second language second language speech learning comparing across the agespeech learning comparing across the age range from childhood into adolescence, range from childhood into adolescence, and into adulthoodand into adulthoodwrtwrt the Critical Period Hypothesis, direct the Critical Period Hypothesis, direct comparison between children at different comparison between children at different ages and adults isages and adults is particularly interestingparticularly interestingiif puberty marks the closing of thef puberty marks the closing of the window window of opportunity for learning, one would of opportunity for learning, one would expect aexpect a discontinuitydiscontinuity in speech learning in speech learning abilities for pre-pubertyabilities for pre-puberty and post-puberty and post-puberty children (and adults)children (and adults)

the experimentthe experiment

the perceptual training procedurethe perceptual training procedure::

thethe learning of lexical tones in Mandarin learning of lexical tones in Mandarin Chinese by AmericanChinese by American adultsadults and childrenand children

MandarinMandarin phonemically distinguishes four phonemically distinguishes four tones differing in pitch height and contour tones differing in pitch height and contour shapeshape

resultsresults

significant improvement in thesignificant improvement in the identification of Mandarin Chinese lexical identification of Mandarin Chinese lexical tonestones

after theafter the 2-week perceptual training2-week perceptual training

for American children at 6, 10,for American children at 6, 10, and 14 and 14 years old, as well as American adultsyears old, as well as American adults

The Perception and Production of The Perception and Production of English /English // and // and // by Korean/ by Korean

Children and Adults Living in North Children and Adults Living in North AmericaAmerica

Kimiko Tsukada, David Birdsong, Ellen Kimiko Tsukada, David Birdsong, Ellen Bialystok, Molly Mack, Hyekyung Sung Bialystok, Molly Mack, Hyekyung Sung

and James Flegeand James Flege

abstractabstract

the production and categorialthe production and categorial discrimination of English /discrimination of English // and // and // by / by native Korean (NK)native Korean (NK) adults and childrenadults and children aa total of 108 participants (72 NK and total of 108 participants (72 NK and 36 36 age-matched native English speakers) age-matched native English speakers) were tested twicewere tested twicetthe NK participants were further he NK participants were further subdivided according tosubdivided according to length of length of residence (LOR) in North Americaresidence (LOR) in North America

iinn Experiment 1Experiment 1 (perception) (perception), both the NK adults , both the NK adults and NK children differedand NK children differed from age-matched from age-matched native English (NE) participants innative English (NE) participants in discriminating discriminating /// from // from ///but the NK children showedbut the NK children showed more accurate more accurate discrimination than the NK adults diddiscrimination than the NK adults did in in ExperimentExperiment 2 (production), 2 (production), the NKthe NK children’s children’s production of the two target vowels closelyproduction of the two target vowels closely resembled NE children’sresembled NE children’shhowever, the NK adults failed toowever, the NK adults failed to produce as produce as large a contrast as NE adultslarge a contrast as NE adults tthe differinghe differing pattern of results obtained for pattern of results obtained for discrimination anddiscrimination and production production for the NK children for the NK children but not the NK adultsbut not the NK adults suggested that the relation suggested that the relation between production andbetween production and perception abilities in a perception abilities in a second language (L2) might dependsecond language (L2) might depend on the age on the age of L2 learningof L2 learning

the experimentsthe experiments

tthehe performance performance of of NK adults and childrenNK adults and children was compared towas compared to age-matched NE speakersage-matched NE speakers

bboth the NK and NEoth the NK and NE participants were tested participants were tested on two occasions separated byon two occasions separated by about 1.2 about 1.2 yearsyears

tthe two English vowels examined here –he two English vowels examined here – /// / and /and // – occur in a portion of vowel space / – occur in a portion of vowel space occupied byoccupied by a single Korean vowel, and are a single Korean vowel, and are of known learning difficultyof known learning difficulty for Koreansfor Koreans

the questions the questions addressed addressed

1.1. Would theWould the NK children perceive and produce NK children perceive and produce the contrasts betweenthe contrasts between English /English // and // and // better / better than NK adults with the samethan NK adults with the same LOR in North LOR in North America?America?

2.2. Would greater differences existWould greater differences exist between NK between NK and NE adults than between NK and NEand NE adults than between NK and NE children?children?

3.3. Would the NK children show greater evidenceWould the NK children show greater evidence of speech learning from Time 1 to Time 2 than of speech learning from Time 1 to Time 2 than the NKthe NK adults?adults?

results Exp.1results Exp.1

ssignificantlyignificantly higher scores were obtained higher scores were obtained for NE adults than NE childrenfor NE adults than NE children

whereas the opposite held true for NK whereas the opposite held true for NK participantsparticipants

KoreanKorean adults with an LOR of 5 years adults with an LOR of 5 years obtained lowerobtained lower scores than Korean scores than Korean children with an LOR of 5 yearschildren with an LOR of 5 years (Child-5)(Child-5)

results Exp.2results Exp.2

tthe NK children produced much largerhe NK children produced much larger contrasts than the NK adultscontrasts than the NK adults

but there was little differencebut there was little difference between NE between NE adults and childrenadults and children

general resultsgeneral results

NK children have learned toNK children have learned to produce and, produce and, to a lesser extent, perceive the distinctionto a lesser extent, perceive the distinction between English /between English // and // and // in a native-like / in a native-like fashionfashion following 3 years of residence in following 3 years of residence in North AmericaNorth America

for for thethe NK children, development of native-NK children, development of native-like productionlike production occurred occurred beforebefore native-like native-like perceptionperception

Relationship between Relationship between LL22production and perceptionproduction and perception

ssome studies suggested that native-likeome studies suggested that native-like perception may precede native-like production perception may precede native-like production ((BordenBorden et al. 1983, Flege 1993) et al. 1983, Flege 1993)

butbut others suggested the opposite others suggested the opposite ((SheldonSheldon et et al. 1982)al. 1982)

iit is generally agreed that the relationship t is generally agreed that the relationship between the production andbetween the production and perception is perception is complex, and is affected by factors such ascomplex, and is affected by factors such as amount of L2 experience amount of L2 experience ((LlisterriLlisterri 1995) 1995)

here…here…tthe NK adults differed from NE adults bothhe NK adults differed from NE adults both in perceiving and producing English /in perceiving and producing English // and / and ////whereas thewhereas the NK children differed from the NK children differed from the NE children only inNE children only in perceptionperceptiontthis pattern might be taken to mean that his pattern might be taken to mean that L2L2 learners will come to learners will come to produceproduce L2 L2 vowels accurately vowels accurately beforebefore developing the developing the ability to ability to perceiveperceive L2 vowels in a L2 vowels in a native-native-like fashionlike fashion

Articulatory and perceptual Articulatory and perceptual influences on the production ofinfluences on the production ofnon-native consonant clustersnon-native consonant clusters

Lisa DavidsonLisa Davidson

abstractabstractoone hypothesis regarding speakers confronted ne hypothesis regarding speakers confronted withwith sequences phonotactically ill-formed for sequences phonotactically ill-formed for their language istheir language is that they will treat these that they will treat these structures as comparably “illegal”structures as comparably “illegal”hhowever, while Englishowever, while English speakers given non-speakers given non-native fricative-initial onset clusters donative fricative-initial onset clusters do not not produce them with perfect accuracyproduce them with perfect accuracy, , they are betterthey are better when the first consonant (C1) of when the first consonant (C1) of the cluster is /f/, followedthe cluster is /f/, followed by /z/, and lastly by /v/ by /z/, and lastly by /v/ tthese findings can be interpreted inhese findings can be interpreted in terms of a terms of a combination of perceptual and articulatorycombination of perceptual and articulatory influences on phonotactic structuresinfluences on phonotactic structures

predictionspredictionsL2 speakers do not produceL2 speakers do not produce all non-native all non-native clusters with equal accuracy, even if all of theclusters with equal accuracy, even if all of the clusters in question are illegal in the speaker’s clusters in question are illegal in the speaker’s nativenative languagelanguage e.g.e.g., an investigation of the production, an investigation of the production of English of English /pr/, /br/, /fr/, initial clusters by Japanese and/pr/, /br/, /fr/, initial clusters by Japanese and Korean speakers,Korean speakers,where neither language has /r/ as thewhere neither language has /r/ as the second second segment of a cluster,segment of a cluster,showed that while the speakersshowed that while the speakers had relatively had relatively little trouble with /pr/, they were less accuratelittle trouble with /pr/, they were less accurate on /br/, followed by /fr/on /br/, followed by /fr/

tthese results were explainedhese results were explained using the using the phonological notion of phonological notion of markednessmarkedness::

as elementsas elements of the clusters became more of the clusters became more marked (voiced stops are moremarked (voiced stops are more marked than marked than voiceless stops but less marked than fricativesvoiceless stops but less marked than fricatives))

L2 learners were less likely to produce them L2 learners were less likely to produce them correctlycorrectly

here:here:

the markedness of certain place and voicethe markedness of certain place and voice features is based on specific phonetic features is based on specific phonetic characteristics thatcharacteristics that distinguish them from the distinguish them from the optimal fricative /s/optimal fricative /s/

experimentexperiment

English speakers were tested on theirEnglish speakers were tested on their ability to ability to produce clusters beginning with /s/, /f/, /z/, and produce clusters beginning with /s/, /f/, /z/, and /v//v/

CzechCzech was was used to record the auditory stimuliused to record the auditory stimuli

ssince English has /s/-initial clusters, it isince English has /s/-initial clusters, it is assumed that being fricative-initial per se is not assumed that being fricative-initial per se is not disalloweddisallowed by the English grammarby the English grammar

it is the place and/orit is the place and/or voicing of /f/, /z/, and /v/ voicing of /f/, /z/, and /v/ that are prohibitedthat are prohibited

the sibilants /s, z/ the sibilants /s, z/ are significantly more perceptible thanare significantly more perceptible than

non-sibilants /f,non-sibilants /f, v/v/

1.1. because the sound source isbecause the sound source is filtered by a front-filtered by a front-cavity resonance for sibilants /s, z/, thecavity resonance for sibilants /s, z/, the amplitude of the frication noise is much greater amplitude of the frication noise is much greater than thatthan that for labiodental fricatives, which have for labiodental fricatives, which have no cavity in front ofno cavity in front of the obstructionthe obstruction

2.2. thisthis high-intensity noise and clear timbre high-intensity noise and clear timbre contribute to thecontribute to the salience and distinctiveness of salience and distinctiveness of /s/, making its acoustic/s/, making its acoustic properties almost properties almost vowel-likevowel-like

3.3. alveolar fricativesalveolar fricatives have a long anterior cavity, have a long anterior cavity, resulting in a well-defined,resulting in a well-defined, distinct spectral distinct spectral shapeshape

4.4. /f, v/ have significantly/f, v/ have significantly lower noise lower noise amplitude relative to /s, z/ amplitude relative to /s, z/

5.5. discriminant analysis and perceptual discriminant analysis and perceptual confusion tests haveconfusion tests have shown that non-shown that non-sibilant fricatives like /f/, /v/, /sibilant fricatives like /f/, /v/, //, and //, and /// are significantly more likely to be are significantly more likely to be confused with oneconfused with one another than are another than are sibilants such as /s/, /z/, /sibilants such as /s/, /z/, // and // and // /

voiced fricatives vs. voiceless fricativesvoiced fricatives vs. voiceless fricatives

voiced fricatives arevoiced fricatives are disadvantaged compared to disadvantaged compared to voiceless ones on articulatoryvoiceless ones on articulatory groundsgrounds::

tthe optimal situation for obstruent voicing occurshe optimal situation for obstruent voicing occurs when oral pressure is maximally when oral pressure is maximally lowerlower than than glottal pressureglottal pressure

ffor the most favorable frication, however, oral or the most favorable frication, however, oral pressurepressure should be maximally should be maximally higherhigher than than atmospheric pressureatmospheric pressure

thus setting up conflicting articulatory thus setting up conflicting articulatory requirements for therequirements for the production of voiced production of voiced fricativesfricatives

vvoiced obstruentoiced obstruent c clusterslustersare further disadvantaged by the fact that they are further disadvantaged by the fact that they areare longer in duration than single voiced longer in duration than single voiced obstruentsobstruentswhich maywhich may require the conflicting air pressure require the conflicting air pressure requirements to berequirements to be sustained for longer than the sustained for longer than the speech system canspeech system can accommodateaccommodatevvoiced fricative-stop clusters haveoiced fricative-stop clusters have articulatory articulatory shortcomings similar to those ofshortcomings similar to those of fricative-fricative fricative-fricative clustersclustersjust as frication requires highjust as frication requires high oral pressure, oral pressure, stops block the flow of airstops block the flow of airwhich likewisewhich likewise causes oral pressure to increasecauses oral pressure to increaseadversely affecting theadversely affecting the pressure drop needed for pressure drop needed for voicingvoicing

the hypothesisthe hypothesis

English speakers fail to accurately English speakers fail to accurately produce non-nativeproduce non-native clusters with initial clusters with initial fricatives other than /s/ becausefricatives other than /s/ because

(i) the(i) the weak-intensity fricatives (like /f/ weak-intensity fricatives (like /f/ and /v/) contain poorand /v/) contain poor perceptual cues, andperceptual cues, and

(ii) voiced obstruent sequences (like(ii) voiced obstruent sequences (like those those beginning with /z/ or /v/) are articulatorily beginning with /z/ or /v/) are articulatorily difficult todifficult to produceproduce

Pseudo-Czech word-initial clustersPseudo-Czech word-initial clusters

/s//s/ /sm/, /sn/, /sf/, /sp/, /st/, /sk//sm/, /sn/, /sf/, /sp/, /st/, /sk/

/f/ /f/ /fm/, /fn/, /fs/, /fp/, /ft/, /fk//fm/, /fn/, /fs/, /fp/, /ft/, /fk/

/z/ /z/ /zm/, /zn/, /zv/, /zb/, /zd/, /zg//zm/, /zn/, /zv/, /zb/, /zd/, /zg/

/v/ /vm/, /vn/, /vz/, /vb/, /vd/, /vg//v/ /vm/, /vn/, /vz/, /vb/, /vd/, /vg/

possible response typespossible response types

resultsresults

/sC/ /sC/ >> /fC/ /fC/ >> /zC/ /zC/ >> /vC/ /vC/tthesehese findings suggest that weak-intensity non-findings suggest that weak-intensity non-sibilant fricativessibilant fricatives are disadvantaged relative to are disadvantaged relative to /s//s/but voice is even morebut voice is even more detrimental to English detrimental to English speakers’ ability to produce thespeakers’ ability to produce the clusterclusteras evidenced by the decreased accuracy on /z/as evidenced by the decreased accuracy on /z/ relative to /f/relative to /f/aas expected, /v/-initial clusters induce thes expected, /v/-initial clusters induce the poorest performance of all the fricativespoorest performance of all the fricativessince /v/ is bothsince /v/ is both non-sibilant (weak intensity) and non-sibilant (weak intensity) and voicedvoiced

performance based on initial fricativeperformance based on initial fricative

error typeserror types

discussiondiscussion

hidden markedness relationshipshidden markedness relationships among /s/, /f/, /z/, and /v/-initial clustersamong /s/, /f/, /z/, and /v/-initial clusters

what is what is their origintheir origin??

acc. to D., acc. to D., threethree possibilities:possibilities:

they result from a they result from a universaluniversal hierarchy hierarchy

theythey are a default relationship present in are a default relationship present in the the initial stateinitial state

theythey are are language-specificlanguage-specific

universal hierarchy?universal hierarchy?

iif the relationships are universal,f the relationships are universal, then all then all languages should respect the hierarchy languages should respect the hierarchy found in thefound in the experimental resultsexperimental results::

/sC/ /sC/ >> /fC/ /fC/ >> /zC/ /zC/ >> /vC/ /vC/

however…however…while languages like Dutch or Norwegian while languages like Dutch or Norwegian allowallow some /f/-initial clusters but no /z/-some /f/-initial clusters but no /z/-initial onesinitial oneslanguageslanguages like Italian and Serbo-Croatian like Italian and Serbo-Croatian have /z/-initial clusters buthave /z/-initial clusters but not /f/-initial not /f/-initial onesonesstillstill, if a language has /v/-initial, if a language has /v/-initial clusters, clusters, like Tsou, Greek, or many Slavic like Tsou, Greek, or many Slavic languages, thenlanguages, then they also have /f/ and /z/-they also have /f/ and /z/-initial ones (note that all of theseinitial ones (note that all of these languages have the phonemes /s/, /f/, /z/, languages have the phonemes /s/, /f/, /z/, and /v/)and /v/)

initial state?initial state?

iif the relationships are present inf the relationships are present in the initial the initial state, but can be modified based on input state, but can be modified based on input fromfrom one’s native languageone’s native language

then speakers of all languages thatthen speakers of all languages that do not do not have /f/, /z/, or /v/-initial clusters should have /f/, /z/, or /v/-initial clusters should perform likeperform like the English speakers on the the English speakers on the production task tested in thisproduction task tested in this experimentexperiment

language-specific relationshipslanguage-specific relationships??

couldcould perhaps result from differences in perhaps result from differences in how the learning processhow the learning process occurs for occurs for distinct languagesdistinct languages

iin this case, significantn this case, significant cross-linguistic cross-linguistic variation on the task examined here wouldvariation on the task examined here would be expectedbe expected