Only S5

mproved inidentificationor discrimination as a func-tion of training (see Appendix D). Improvement in iden-tificationby e

Finally, the results of minimal-pairs tests showedpositivetransferto word-initial/r/versus /1/ for onlytwo subjects, including S5, who hadperformedmostpoorly on training andposttestsofsyntheticseries

consistedofanidentificationfunctioninwhichall10 stimuli were labeled with less than 80%accuracyand discrimination was not above 58% correctfor any comparison pair.

like-wise, general claimsaboutthe malleabilityofphoneticperceptual processes in adulthood cannot be made onthe basis of one or two specific instances

development of selective percep-tual skills that become automatic, suchthatphonologi-cally relevantphoneticparameters areabstractedanddifferentiated,whereas nondistinctive variations arefiltered out and ignored

Performance on/w/-/r/

one or more phonetic variantsthatare similar to those present in the native languagemay be easier than learning a contrast in whichbothcategories are phonetically dissimilar to any nativephoneme

The Japanese were required tolearn adistinctionbetween two phonemes, neither ofwhich are similar phonetically to any phoneme in theirlanguag

Futurestudies

We can speculate that subjectsmay actually havedifferentiatedthree categories duringidentificationposttests,but were unable toreportthisbecause of therestrictionto two response label

Theirposttestperformance remained incon-sistent for both trained-on and transfer stimulus materials

Sub-jects S2

regardless of their pretraining perfor-mance level

one canreadily see the positive effects of training on subsequenttests ofperceptionof synthetic seriescontrasting/r/and /1

straightforward

identificationfunctions showed highlyconsistent labeling (>90%) of all but the two stimuliclosest to the categoryboundaryand discriminationfunctions had peaksofmost accurate discrimination forcross-category comparison pairs

no majornonmonotonicit

identificationand discrimi-nationfunctions for which performance was above thechance level (p<.01) for some stimuli,butthe identi-ficationfunctionwasnonmonotonicand discriminationpeaks were not predictable fromidentificationcategoryboundaries

they came to discriminate as well as orbetterthanthe Americans in tqeir pretraining perfor-mance

Althoughsubjects showed considerable vari-ability, all eight subjects improved with training

gradualimprovementover sessions with thegreatestimprovementin the first several session

Althoughthere wereconsiderableindividual differ-ences inperformanceon thepretests,the foursubjectsreceivingtrainingfirst(SI-24,henceforthlabeledtheEgroup)didnotdiffer from thefourcontrol(C) sub-jects(S5-S8) as a group. However, this is because Sl andS4 were the best onpretestperception,whereas S2 andS3 were among the worst

llsubjectswere given theidentificationanddiscriminationtestsoftherock-lockand rake-lake series and the minimal-pairs tests

Eachsubjectthencompleted14 to 18trainingsessions, whichtookplace on separate days over the course ofabout3 weeks

Each subject was tested individually during train-ing sessions. The stimuli werepresentedbinaurallyover ear-phones

dentificationandodditydiscriminationtests wereconstructedseparatelyfor the rock-lock and rake-lakesyntheticseries

Intonationcontourswereidenticalfor all 10stimuli

utilizedin previousstudiesand aredescribedindetailin theoriginalreferences

difficultyin perceiving andproducingIIIandIII

Eight female native speakers ofJapanesewererecruitedfrom a weeklyintermediate-levelEnglish-as-a-second-languageclasssponsoredby a foreignstudentsorganizationofthe Uni-versity of Minnesota.

trans-feroftrainingtoperceptionofnaturalspeechstimuli

fourothersubjectsserved as anindependentcontrol group

Four subjects received training immediatelyafter completing pretests

pretraining versus posttraining tests

intensive con-versationalinstructionandproportionallyhigh everydayuse of Englishproducedvery consistentidentificationfunctions

differentiating these phonemes in AEnaturalspeechminimal pairs

/r-l/contrastis not distinctive in Japanesephonology,

How easily does a change occur?What is the nature of the change? Does improvement inthe training task with one set of stimuli transfer toother tasks andotherstimuli? Does the change in per-ception ofsyntheticspeech series transfer to tests of/r-l/ perception using real speech?

directly study possible transfer of training effects to theperception of foreign contrasts in natural speech

intensive conversationalinstruction(Withnative speakers) iscorrelatedwith improvedperceptionofthe foreign contrast

Americans had difficultydifferentiatingsyntheticoraland nasal vowels/ba/-/ba/,while Hindi speakers, forwhomthe contrast is distinctive, perceived thecontrastcategorically.

learning a newpatternofspeechproductionin a foreign language is problematic.

NativeJapanesespeakerslearningEnglishhavedifficultyperceptuallydifferentiatingtheliquidconsonantsIrlandIll,evenafterextensiveconversationalinstruction

he neurocognitive approach and the novel word learning paradigm

ntegrated brain network

nformation processing

SL learnersand LSL learnersrecruit different networks and hubs tohandle the same lexical task

One important area for phonological processing, the left posterior superior temporal gyrus (BA 22),also showed less activation at T2 than at T1 for our learner

despite the lack of differences between learners and non-learners in their behavioral performance, their brain activation patterns differed significantly

The purpose of the present study was to examine neurocognitive signatures of successful learningof new vocabulary in a second languag

successful learners,compared to less successful learners, performed more accurately in all three conditions of the sounddiscrimination task at T2

did notcomplete the study across the six-week period

uSEM could be used to examine contemporaneous and lagged (sequentially) relationshipsbetween ROIs in a blocked-fMRI study, whereas euSEM is used for data from er-fMRI studies

uccessful learners (SL) group and less successful learners (LSL) group

T1 equilibration

MRI images

wordepicture association judgment task

T and O discrimination task

xperimental condition

ound discriminationexperiment during the pre-training fMRI scan, and then six weeks later the post-training fMRI scan

Their response accuracy in the recognition task during thetest phase of each training session was used to indicate their L2 learning success after that trainingsession.

participants completed a recall testas well as a recognition test without feedback

judged to be perceptually natural andaccurate by four native Mandarin speakers

6 monosyllables of the CVC structure

The learners underwent 18 training sessions(three per week) in six weeks and learned 48 Chinese pseudowords

native speakers of Englis

Thirty-nine right-handed (Snyder&Harris, 1993) healthy adults

brain connectivity

Thus, we will perform“effective connectivity”analyses ofour data

uSEM and euSEM

functional changes in the brain that occur in the L2 word learners.

identification of the direction of influences

first goal

thesuccessful learners' network, on average, had reduced local efficiency but increased global efficiency ascompared with that of the less successful learners, suggesting that a more cost-efficient network or-ganization underlies sound-to-word learning abilities for the better learners

between brain localization and brain organization, and will undoubtedly lead to newinsights into the bilingual brain in the years to come.

learning of a second language (L2) in general is much more difficult and often less effective ascompared to learning of afirst language (L1), especially when the learning takes place later in life.

But whatchanges are taking place in the brain as learning progresses? Howcan we identify differences in brain changes that reflect successesof learning?

neuroimagingmethods especially functional magnetic resonance imaging (fMRI

The evidence so far suggests that bilingual learners have enhanced neural activity involving thiscontrol network for the effective use of the target languages while inhibiting the unintended language.Importantly, this enhanced activity is modulated by the degree of L2 proficiency: increased proficiencytends to lead to decreased brain activity, especially in the prefrontal cortex, indicating a more efficientuse of the control network for language by the proficient bilingual

understanding distinct vs.overlapping neural systems

dentify individual differences by examining the functional neuralcorrelates associated with the learning of novel L2 words before and after training

surge of interest in neurocognitive studies

native Japanese speakers

These data again also suggested thatit is possible to use patterns of neural activity to differentiate good learners from poor learners

recruited twenty-four native Chinese speakers

tronger left posterior STG activation

rightSTGandrightIF

classified

a real word-learning task

native English speakers to learn Hindi dental-retroflex contrasts

both the expansion of preexisting language-related areas

six adult native speakers of English

increased tone-identification performance was associatedwith an increase in the spatial extent of activation in left superior temporal gyrus (

short-term language training studies

39 native English speakers who had no prior knowl-edge of Chinese or other tonal language

high-variability auditory training