Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
gp± fiff fi rg lg geg 3 e2015(ilZEk 27) tP l2 fi2436.g.
Joumal ofthe Phonetic Society
ofJapan, Xlo1. 19 No. 3 December 2015, pp. 24-36
msr.gA"Critical Period" in Seeond Language Acquisition Reconsidered:
Production Patterns and Listening/Reading
Proficiencies of Japanese Learners
Michiko MocHizuKi SuDo* and Ikuyo KANEKo*
ag=s- ge. apec:-HtsK\ygotnt/tgtsct6annmeftsxl6->tagash
・ ftashoen.6..hs6-
SUMMARY: The present study investigated the effect of living in the target language country and that ofthe age factor
on the acquisition ofEng]ish by Japanese learners. Three subject groups panicipated in this study: Native speakers of
American English, Japanese returnees who had lived in the U.S., and cQllege students who had never lived overseas. For
measuring listening and reading abilities, we employed orncial TOEICsc scores. We also examined two factors ofdura-
tional control: ISI durations and stressed vowel shortening. The results showed significant difTerences in all parametersexamined between the returnees and general college students, We also observed that the age factor had an effect on the
different degrees of attainment efthe production proficiency and on 1istening ability ofthe returnees, but not on their read-
ing ability.
Key words: second language acquisition, age factor, criticaL period,compensatory vowel shortening, ISI
1. Introduction
A variety of factors have been reported in the lit-erature as having influence on second language (L2)acquisition (Altman 1980, Larsen-Freeman and Long1991, Skehan 1989). The individual learner variablesinclude age, aptitucle, motivation, attitude, personality}cogriitive factors, hemisphere specialization, learningstrategies, memory, awareness, will, proficiency in thefirst language (L1), interest, sex, birth order, and priorexperiences. Among them, age is one of the factorsthat has received the most attention in the field of L2acquisition (Ellis 1994). Age of initiation of languageacquisition can be regarded as a crucial variable to
distinguish between Ll and L2 acquisition. In lan-
guage acquisition, there is a general belief that youngerleamers do better than older learners. This belief isbased on the concept ofa critical period. It refers to a
biologically constrained peried in which language canbe acquired more easily than at any other time, and
beyond which complete mastery of a fbreign language
is no longer possible. This concept of the critical pe-riod was originally introduced by Penfield and Roberts
(1959), positing that children have a specific capacity
fbr language learning due to cerebral fiexibility that
subsides at appreximately age 9, Lenneberg (1967), themost well-known researcher for claiming the CriticalPeriod Hypothesis (CPH), proposed a critical periodstarting from about age 2 and continuing to puberty,He also suggested that language learning may be moTediMcult after puberty because the brain lacks the abil-ity fbr adaptation. The theory that the learner's age isa significant factor in restricting language acquisitionhas also been described as the notion of a
"Sensitive
Period" or "Maturational
Constraints" (Bornstein 1989,Knudsen 2004, Lamendella 1977, Long 1990, Oyama
1976, Snow 1987).
[Itiggered by Lenneberg's writing (1967), the ideaof the critical period was extended to the field of L2acquisition. A number of studies have claimed that theage factor affects L2 acquisition (Asher and Garcia1969, Bornstein 1989, DeKeyser 2000, Long 1990,
Oyama 1976, Scoyel 1981, Singleton 2005, Snow1987, Tahta, Wbod and Loewenthal 1981), supportingthe CPH. Major (2001) claimed that both the leanner7sage of arrival in the L2-speaking country and the age
of Iearning are important variables governing whethegor to what degree, a learner can attain native-like
' School of Health and Sports Science,Juntendo University (MfiKEjltl7Z pt - 7 reptN?gB)
-24-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
A "Critical
Period'' in Second Language Acquisition Reconsidered
pronunciation, Several studies have observed the difi
ferences in L2 perception and produetion between
early learners and late leamers. McCarthy, Evans and
Mahon (2013) investigated the production of stops
and vewels of Sylheti (Ll) and English (L2) by adult
speakers who had been residents in the UK fbr similar
lengths of time but had arrived there at different ages.
Analyzing Vbice Onset Time (VOT) of stops and vowel
formants and duration, they reported that the late arriv-
als (speakers who had arriyed in the UK after age 16,
mean age of anival: 21) produced native-iike Sylheti
stops and vowels, but that their English categories were
influenced by their Ll. 1[1ie early arrivals (speakerswho had arrived in the UK before the age of 16, mean
age of arrival: 6) and the second-generation speakers
(those who were bom in the UK) produced native-like
Sylheti vowels but not Sylheti stops, while their Eng-lish production was similar to the Standard SouthernBritish English. Darcy and Kruger (2O12) examined the
early L2 learners and investigated the age at which Ll
vowel categories influence the acquisitien ofL2 vowels
in ten-year-old Turkish children acquiring Gerrnan as
their L2. [Ihey tested 28 children (14 native speakers
of Germari and 14 Turkish-German bilinguals) by both
perception and production experiments. In the percep-tion experiment, they tested the children's ability to
discriminate four German vowel contrasts using an
oddity task while they examined their production of
the same vowels with an un-cued word naming task
in the production experiment. Tlie results showed that
the TurkishGerman bilingual children categorized
perceptually dificult German contrasts difTerentlyfrorn German monolingual children, In contrast to the
perception results, the two groups of children were
globally indistinguishable in production, Several other studies, on the other hand, haye shownthat the production of consonants by 1ate L2 learners
displayed their L1 features (Flege, Munro and MacKay1995, Sundara, Polka and Baum 2006). Likewise,the late L2 learners showed the Ll features in the L2vowel production (Baker and Trofimovich 2005, Flege,MacKay and Meador 1999). Sudo and Kaneke (2012)reported the differences in the production ef English
rhythmic patterns between early L2 learners (whoseage ofarTival was before puberty) and latc L2 leamers
(whose age of arrival was after puberty) by measuring
durations of interstress intervals (ISI), suggesting a
different clurational control of stressed and unstressed
syllables in the ISI by the two subject groups. It shouldbe noted that this previous study does not assume equal
ISIs in English. Quite a few studies have shown that
perfect isochronM on the level of production, does not
exist (Beckman 1992, Ericksson 1991, Laver 1994,Lehiste 1977 for review). On the basis of acoustic
measurements, some researchers have rejected the
dichotomy between stress-timed and syllal)le- or mora-
timed languages (Dauer 1983), Others haye reported a
weak categorical distinction between stress-timed and
syllable-timed languages, based on durational variabil-
ity (Grabe and Low 2002). Evidence fbr the compensa-tory control in ISIs has also been presented in varieusstudies from the viewpoints ofboth production and per-ception (Fowler 1977, Huggins 1975, Mochizuki-Sudo
and Kiritani 1991, Rakerd, Sennett and Fowler 1987).The results of the above-mentioned studies indicatethat ISI is a rhythrnic unit in the production and percep-tion of English. Lending support to this claim, Sudo
and Kaneko (2012) observed that the early L2 learners
shewed a significant amount of approximation to the
production patterns of native speakers of English in
terms of the normalized ISI durations and the incre-ments of increase in the normalized ISI durations,while no statistical differences in the ISI durational
patterns were fbund between the late L2 learners and
Japanese learners of English who had neither livednor studied abroad, This result could be interpreted as
indicating that the factor of age of arriyal has an effect
on different degrees of L2 attainment from the view-
point of the production proficiency. Likewise, early L2learners displayed production patterns that are similar
to that ofnative speakers ofboth languages, L1 and L2
(Baker and liofimovich 2005, Guion 2003). Regarding
production, there are some studies which have shown
that even late L2 learners are successfu1 learners in thatthey have become indistinguishable from monolinguals
(Bongaerts, Planken and Schils 1995, Kinsella 2009,
Neufeld 1977, 1979, 1980). In sum, significant early-
late differences have been reported fbr L2 perception,and diflerent degrees of early-late effects have been
observed for L2 production studies,
In order to account for age effects on speech perfbr-
mance, various explanations have been of:ered in L2
research. The most common, and the biological expla-
nation is the critical period concept, as we have men-
tioned above. Another type of explanation is that age
effects arise from input diffbrences between Ll and L2
(Oyama 1976, Snow and Hoeihagel-Hohle 1977, Sudoand Kaneko 2005). What this explanation indicatesis that the accuracy of L2 pronunciation depends onhew much native-speaker input as opposed to fbreign-accented L2 input is received (Flege and Liu 2001), Inother words, late L2 learners receive less adequate L2
-25-
NII-Electionic
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
ffZP. 3t(ResearchArticles)
phonetic input than early learners. The phonetic inter-action between L1 and L2 can be offered as another ac-
count for age effects. The Speech Learning Model pos-ited by Flege (Flege 1995, 1999, 2002, Flege, Schirruand MacKay 2003) suggests that the phonic elementsmaking up the Ll and L2 phonetic sub-systems of a
bilingual exist in a common phonologica] space, rather
than in twe separate phonological systems, and the
two phonetic sub-systems interact each other. Anotherexplanation, the Perceptual Assimilation Model (Best,McRoberts and Goodell 200 1 ) indicates that interactionbetween L1 and L2 perception increases with age.
Whatever the source ofan age factor is, this specificfactor in L2 acquisition seems to be undeniable, Thedifferences in the age effect among various studies
suggest that the degrees of attainment of L2 acquisition
might vary, depending on the aspects of acquisition.
In other words, the age factor has different degreesof effects on various features of L2 production and
perceptlon.
Regarding another aspect of production in English,a cQmpensatory vowel shortening which relates to
the production of English rhythmic patterris has beenreported in various studies (Fant, Kruckenberg and
Nord 1991, Fowler 1977, Huggins I975, Katz 2012,
Munhall, Fowler, Hawkins and Saltzman 1992, Rakerd,Sennett and Fowler 1987, Shaiman 2001). This phe-nomenen refers to a shortening of stressed vowels with
the addition of fbllowing unstressed syllables in theISI, Fowler (1977) reported that the average percent-age ofthe stressed vowels from a one-syllable ISI to atwo-syllable ISI was 12%, ranging from 6 to 18% forsix sentence sets. Mochizuki-Sudo and Kiritani (1991)reported a foet-level shortening for the native speak-
ers of English, observing that the average percentageof stressed vowel shortening from a one-syllable ISIto a twe-syllable ISI was 19%, while Japanese non-
profieient speakers of English displayed very littlefoot-level shortening; an average of 7% shortening.
Sudo, Kaneko and Nishimura (2012, 2013), and Sudoand Kaneko (2013) analyzed the compensatory vowel
shortening by Japanese learners of English and exarn-
ined their English proficiency in terms ofthe durationofstressed vowels in ISIs. The results showed that sig-
nificant difierences were observed in the cornpensatoryvowel shortening between Japanese learners of Englishwho had studied English in Japan and Japanese return-ees who had lived in the U.S.
In order to measure the proficiency of English as L2,the fest of English fbr International Communicatien
(TOEICop) is one ofthe most globally adopted standard-
ized tests. It is reported to provide a valid assessment
of English-language reading and listening proficien-cies (In'nami and Koizumi 2012, Powers 201O), It is amultiple-choice assessment consisting ofListening and
Reading Sections. Vhrious factors related to TOEICscores have been reported in literature. Tb cite a few,Sudo, Mochizuki, Itoh and Kirino (2011) reperted sig-
nificant relationships between the L2 reading proficien-cy and the Ll working memory in addition to the sex
diflerences in L2 acquisition, They also reported that
TOEIC scores and reading rate in English exhibited a
posjtive correlation with the working memory-related
parameters in the female group, but not in the male
greup, Mochizuki, Sudo, Itoh and Kirino (2013) inves-tigated the interrelationships among physical abilityl
social functioning, and L2 proficiency and conducted
correlation analyses fbr physical parameters, capacities
of social cognition, and TOEIC scores together with
English reading rate, as well as vecabulary size. Amongthe results, TOEIC seores showed a strong correlation
with reading rate and some correlation with vocabulary,
Also, vocabulary ability showed a negative correla-
tion with one of the social parameters. Regarding theTOEIC sceres, Sudo and Kaneko (2012) showed that
Japanese general college students and returnees were
significantly different in the total TOEIC scores and
those in both Listening Section and Reading Sectien.Thus, TOEIC can be used as one ofthe indices to mea-sure L2 proficiency. The purpose ofthis study was to investigate the age
factor in the acquisition of English by Japanese learn-ers, SpecificallM we attempted to find out the effbcts of
age of arrival in the target language country together
with length of residence on English proficiency of
Japanese Iearners, We analyzed English profieiencyfrom the viewpoints ofproduction patterns and TOEICscores. We focused on the two factors of productionpatterns: ISI durations and stressed vowel shortening.
Based on the results of our previous studies, ISI dura-tional control and stressed vowel shortening are shown
to be the parameters ofnativelikeness. Using TOEICscores, we measured listening and reading abilities.
We were interested in finding out the efTbcts of an age
factor on production patterns and listeninglreading pro-ficiencies. Since speaking ancl listening abilities were
reported to have high corre]ations (Bozorgian 2012),
we hypothesized that there are relationships between
production patterns and listening proficiencies. We
were also interested in the relationships between theEnglish proficiency measured by TOEIC scores and
the proficiency measured by production patterns. The
-26-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
A`LCriticalPeriod" in Secend Language Acquisition Reconsidcred
research questions addressed in the present paper are:
1) Does an age factor have an effect on various degrees
ofL2 attainment, depending on different features of
production and perception?2) Is there a
"critical period" in the acquisition of Eng-
lish by Japanese learners?In this study, we attempted to find out the effects of
an age factor on different aspects of production and
perception, testing the hypothesis ofa "critical
period"
in L2 acquisition.
Thble1 Age ofarrival and length ofresidence for RTN.
Age ofarriyal
Lengthofresidence
RTN3 (n ==11) RTN4(n=13)
RTN] (n=17)
RTN2 (n=7)
101 76
2. Methods
2.1 Subjects
We prepared three subject groups for the produe-tion experiment: 13 adult native speakers ofAmerican
English (AMR: 8 males and 5 females), 24 Japanese
retumees who had lived in the U,S, (RTN: 12 males
and 12 females), and 13 Japanese learners of English
who had neither lived nor studied abroad (JPN: 9 malesand 4 females), All subjects from MN started theirEnglish study when they entered a junior high school.
The Japanese learners were all ¢ ollege students who
belong to either a School ofHealth and Sports Science
or a Scheol ofMedicine, and teok the oMcial TOEIC
test fbr the measurement oftheir English proficiency,
The age of arrival in the U,S. fbr the subjects of
RTN was O to 1 8. We divided RTN into two subgroups
according to the age of arrival in the U.S. The age of ar-
rival fbr RTNI was O to 9; the age of anival for RTN2
was 13 to 18 (RTNI: 17 subjects, RTN2: 7 subjects).
We intentionally excluded the subjects from our analy-
sis whose age of arrival was between 10 to 12, since
the definition of puberty varies frorn study to study,
thus making the age of9 and 13 years cut-off points in
this study. Thus, the subjects of RTNI were supposed
to be the subjects befbre the critical period, and those
of RTTN2, the subjects after the critical period, The sub-
jeets of RTN had been in the U.S. fbr over one year at
least, and 12 years at the longest. We also divided the
same RTN into two subgroups according to the length
of residence in the U.S. The length of residence fbr
RTN3 was six years or more (mean=9.3 years); the
length of residence for RTN4 was one to three years
(mean= 1.5 years). Based on the data collected in this
study and to make two subgroups of short versus long
length of residence, we chose three years of length
ofresidence as the cut-off point(RJI]N3: 11 subjects,
RTN4: 13 subjects), The age of arrival and the length
of residence for the subgroups of RTN are presented in
Table l.
Note, RTN=a Japanese retumee who had lived in the U.S.;
RTN1=a returnee whose age ofarrival was O to 9; RTN2=a
returnee whose age ofaniva] was 13 to 18; RTN3=a returnee
whose length ofresidence was six to 12 years; RTN4=a re-
turnee whese length ofresidence was one to three years,
2.2 Materials
Five English sentences were devised as the linguis-
tic material for the production experiment (Tbble 2),
The sentences differed in the number of nominally
unstressed syllables that intervened between a target
stressed syllable and the next stressed syllable. The
number of unstressed syllables ranged frem one to
three. These sentences contained the ISIs within and
between words (ISII), except fbr the one-syllable ISI"Pete."
While the number ofsyllables in the ISI1 varied
from sentence to sentence; the number of syllables in
the fbllowing ISI (ISI2) was always twe syllables, As
shown in Tlable 2, the five sentences contained a target
stressed vowel [i],
Table2 Lingllistic materials for the production experi-
ment.
Number of syllables
in a target ISI1
1, P/etepl/aysthep1iano.
ISI I ISI 2
2, Ptetecanplfaythep1iano,
ISrl ISI2
3. P!eterpl/cuts thepfiano.
ISI I ISI 2
4. P!etercanpt/aythep1iano.
ISI I ISI 2
5. P/etexsonpl1ays thep/iano,
ISI I ISI 2
1
2
2
3
3
2.3
waProcedures measured the Japanese learners' English profi-
-27-
NII-Electionic
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
blnge.x(ResearchArtieles)
Tbble 3TOEIC average scores (JPN & RTN).
Subject group Tbtal (SD) Listening(SD) Reading(SD)
JPN
RTN
MeanRangeMeanRange 422 (61)
]** 241 (24)
]**
335-505
844 (le3)
630 - 985
210-280460
(35)
390-495
iiil:lf,:25iO]
'"
235-490
**p<.Ol
Table 4TOEIC average sce res (RTN1 & RTN2).
Subiectgroup Tota] (SD) Listening(SD) Reading (SD)
RTN1
RTTsl2
MeanRangcMeanRange 867 (90)
710-98S
789(12I)
630-955
319Ill
`
l419cr390 - 465
396 (68)
265 - 490354
(95)
235-490
** p<.Ol
ciency by the oMcial TOEIC test, We also carried out
production experiments to investigate the leamers'
proficiency in production of English sentences. We
held one session fbr each subject, and the subjects were
recorded individually. The utterance list of 25 sen-
tences was arrangecl in a pseudo-random order. Befbre
the recording, they were provided suMcient time forpractice. The subjects were instructed to produce theirbest English-like utterance at a comfbrtable speaking
rate and to read each sentenee through without pausing,wuen they made a mistake, they were asked to repeat
the sentence frorn the very beginning. For each subject,
three repetitions of each sentence were selected fromthe total of five repetitions uttered. Spectrograms andwave forms were made from these recordings. We car-
ried out acoustic analysis of the linguistic materials,
measuring durations oftarget ISIs (ISI1) and fbllewingISIs (ISI2). The duration of ISII was defined as the
interval between the onset of the vowel [i] in "Pete",
"Peter",
and `[Peterson"
and that of the next stressed
vowel in "play", while that of ISI2 was defined as the
interval between the onsets ofthe vowel in "play"
and
the next vowel. For the purpose ofthe normalization of
the speaking rates, the ratio f\S was obtained. Further-more, we measured the durations of the target stressed
vowel, [i], The onset of the vowels was defined as theinstant a sharp rise appeared in the power of the firstfbrmant. Since the target vowel was fbllowed by a stop
[t] or a flap [f], the offSet of the yowel was defined as
the abrupt decrease in power corresponding to the stop
closure or the change in power and formant structurecorresponding to the flap.
3. Results
As shown in Tbble 3, the English proficiencies ofthe
Japanese learners were measured by the oMcial TOEIC
test, The TOEIC average score ofJPN was 422 (Listen-ing Section: 241, Reading Section: 181), The average
score of the Japanese learners who had lived in the U.S.
(RTN) was 844 (Listening Section: 460, Reading Sec-tioni 384). RTN showed rnuch higher scores than JPNin both the total scores and the scores of the Listeningand Reading Sections. Ilie results of analysis of vari-
ance showed that the differences in the total TOEIC
scores (F<1, 35)=181.06,p<.Ol, ij2=,84) and those inboth the Listening Section (F<1, 35)=415.39, p<,Ol,op2=.92)andReadingSection(F<1,35)=75,99,p<.Ol,ny2=,68) between JPN and IU]N were statistically sig-
nificant.
In erder to investigate the effect of the age of arriv-
al-a factor related to the Critical Period Hypothesis
(CPH) in the field of L2 acquisition, we divided iU'Ninto two subgroups according to the age of arrival
in the U,S.: RTNI and RTN2. As shown in Tbble 4,
the TOEIC average score of RTNI was 867 (Listen-ing Section: 471, Reading Section: 396), The aver-
age score of RTN2 was 789 (Listening Sectioni 43S,
-28-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety ofJapan
A `
℃ritical Period" in Second Language Acquisition Reconsidered
[fable5 TOEICaveragescores(RTN3&RTN4).
Subject group Tbtal (SD) Listening(SD) Reading(SD)
RTN3
RTN4
MeanRangeMeanRange 887 (97)
710-985808
(98)
630-965
474 (31)39S-49S450
(35)390-49S
414 (76)
265-490358
(71)
235-470
Reading Section: 354). RTNI showed higher average
seores in the total scores and the scores ofthe two sec-
tions than R[[:N2. The results of analysis of variance
showed that the differences in the total TOEIC scores
(F<2, 34)=99,98,p<.Ol, op2==.85) and those in both theListening Section (F<2, 34)=252.94, p<,Ol, n2=:,94)and Reading Section (F<2, 34)=39.95,p<.Ol, ij2=.70)among RTN1, RTN2 and JPN were statistically signifi-
cant, Furthermore, in order to find the source ofsignifi-
cant differences in the subject groups, multiple compar-
ison te sts (Tukey-Kramer) were carried out. The results
showed that the differences in the tota1 score and the
score of each section between JPN and RTN1and those
between JPN and RTN2 were statistieally significant
(p<.Ol). The differences in the scores ofthe Listening
Section between RTNI and RTN2 were statistically
signifieant Cp<,05), while the differences in the totalscores and those in the Reading Section between these
subject groups were not.
Next, we divided the sarne RTN subjects into twosubgroups according to the length of residence in thetarget language country. Table 5 shows that the TOEIC
average score of RTN3 was 887 (Listening Section:474, Reading Section: 414). The average score of
RTN4 was 808 (Listening Section: 450, Reading Sec-tion: 358), The ayerage tota1 scores ofRT[N3 in additionto the average scores ofListening and Reading Sectionswere higher than those of RTN4. Concretely, there
were eight RTN3s out of 1 1 subjects who scored higherthan 800, while eight RTN4s out of 13 subjects did so.Although RTN3 showed higher average scores than
RTN4, the results ofmultiple comparison tests (Tukey-Kramer) showed that these differences between RTN3and RTN4 were not statistically significant,
Tuming to the production of English sentences bythe same subjects, we measured the durations of target
ISIs (ISI1) and fbllowing ISIs (ISI2), as explained inProcedures. For the purpose of the normalization of
the speaking rates, we obtained the ratio rsrsil,Figure 1shows the results for values ofthe rsrsi'l within a word
produced by the three subject groups: AMR, RTN, and
150%
At20%gV
90%8E
6o%pn
30%o}t
Number of syllables in ISII
Figure 1 DuratiensofISIwithinaword(JRi,S-)forAMR,
JPN, and RTN.
JPN (Sentences 1, 3, and 5 in [Ibble 2).
In Figure 1 , we could observe a durational tendency
related to each ofthe subject groups. The increment ofincrease in the normalized ISI durations with the ad-
ditional unstressed syllables differed among the three
groups of subjects, RT[N showed closer durational pat-terns to those ofAMR. The normalized ISI durationsincreased by the largest increment for JPN, by a smaller
incrernent fbrAMR, and by the smallest increment for
RTN. The results ofanalysis ofvariance for the ISI du-rations within a word ( fi'I ) showed that the differencesin the normalized ISI durations among JPN, RTN, and
AMR were statistically significant, regardless of the
number of syllables in lSII (l syllable: F(2, 147)=7.40,
p<.Ol, n2==.09; 2 syllables: F(2, 147)==5.85, p<.Ol,e2=.07; 3 syllables: F(2, 147)=5.16,p<,Ol, n2=.07),Theresultsofmultiplecomparisontests(Tukey-Kram-er) fbr the ISI durations within a word (-fi,S-2 ) showedthat the differences in the normalized ISI durationsbetween JPN and AMR were statistically significant,
regardless of the number of syllables in ISII (p<.Ol>,The differences between RTN and AMR were shown
to be statistically significant only when ISII was one
syllable (p<,Ol), but not significant in the case of
two syllables and three syllables. The differences indurational pattems in terms of fff2i between JPN and
-29-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
EJffiinsc(ResearchArticles)
150%
120%AS
90%・:ti
60%:a
3o%o%
ijpc,Ol..[.Ol
ijp(,Olitpc,Ol"pc.Ol
'
'1
MRRTNRTrvnN"MRRTNRTNJPN
12L12
11
2
AMRRTNRTNJPN
123
Nllmber ofsyllables in ISII
Fig"re 2 Durations oflSI within aword ( ,rs.'S ) forAMR, JPN, RTNI, and RTN2.
tso%
120%Agv
ge%=・s-E
6oza,=a
30%o%
"*p<,Ol
*ip<,Ol
-p<.Ol e*p<.e]
i#p<.OL
/t
:1/1
4
t
/
[iI'
MRRTNRTTS'JPN
341
MRRTNRTNJ?N
32
MRRTNRTNJPN
343
Number of syllables in ISII
Figllre 3 Durations of ISI within a word ( irs,if ) forAMR,
JPN, RTN3, and RTN4.
RTN were statistically significant when ISI1 was three
syllables Cp<.Q5)・ Regarding the ISIs between words, AMR and RTNshowed similar duratiena] patterns when the ISIs werebetween words. The results of analysis of variance
showed significant difTerences among JPN, RTN, and
AMR in the case oftwo syllables and three syllables
(2 syllables: F(2, 147)=9,46, p<.Ol, ij2=.11; 3 syl-
lables: F(2, 147)=13.07,p<.Ol, n2=.15). The results
of multiple comparison te$t$ (Tukey-Kramer) showed
that the differences between AMR and RTN were not
statistically significant, while the difTerences betweenJPN and IU'N were statistically significant in the caseof two syllables and three syllables (p<,Ol). Also, thedifferences between AMR and JPN were statistically
sigriificant when iSI1 was two syllables and three syl-
lables ip<.O1). In order to investigate the effect ofthe age of arriv-
al-a factor related to the Critical Period Hypothesison the production proficiency, we compared the dura-tional patterns for RTN1 and RTTN2. Figure 2 shows the
durations of ISI within a word ( kYI ) fbr AMR, JPN,
RTN1, and RTN2. In this figure, we could observe that
AMR and RTNI showed similar durational patterns,while the patterns for JPN and RTN2 were shown to
be similar, The results of analysis of variance showed
significant differences in the durations of ISI within
a word (rsrsi'S) among JPN, RTNI, RT:N2, and AMR
(1 syllable: F(3, 143)=-10.26, p<.Ol, n2-.18; 2 syl-
lables: K3, 139)=13,66,p<.Ol, ny2=,23; 3 syllables:I<3, 144)=5.98, p<.Ol, ny2=.11). The results of mul-
tiple comparison tests (Tukey-Kramer) showed that
the differences in -fEfF, between RTN2 and AMR were
statistically significant in the case ofone and two sylla-
bles Ct)<,Ol). The difTerences in i",;ii between JPN and
RTNI were statistically significant when ISII was two
syllables and three syllables (p<.Ol). The differencesbetween JPN and RTN2 were shown to be statistieally
not significant The differences in ,L,S',l between IUrNland RTN2 were statistically significant when ISII wasthree syllables Cp<,O1). The values of
'f:;-, fbr the ISIs
between words showed a similar tendency among the
subject groups, which we observed in the normalizeddurations within words. The results of analysis ofvari-
ance showed significant differences in the durations ofISI between words ( ,rs,I2" ) among these subject groups (2syllables: F(3, 145)=4.54,p<.Ol, ny2=.09; 3 syllables:F<3, 144)=9,Ol, p<.Ol, n2=.16). The results of mul-
tiple comparison tests (Tukey-Kramer) showed signifi-
cant differences in fstSJi between JPN and RT[Nl when
ISII was two syllables and three syllables (p<,Ol). Next, we attempted to observe the effect oflength of
residence in the target lariguage country on production
proficiency, and compared the values of -X,IS-2
for RTN3and RTN4. Figure 3 shows the values of -.2 for ISIwithin a word forAMR, JPN, RTN3, and RTN4, In Fig-ure 3, we observe that AMR and RTN3 showed similar
durationa] patterns, while the patterns fbr RTN4 were
relatively eloser to those ofAMR as compared to these
of JPN. The results of analysis of variance showed
sigtiificant differences among these subject groups(1 syllable: FK3, 143)=12.29, p<.Ol, n2=20; 2 syl-lables: F<3, 138)=13.65, p<.Ol, n2=:.23; 3 syllables:
]F<3, 140)=17,88,p<.Ol, rp2=,28). The results of mul-
tiple comparison tests (Tukey-Kramer) showed that
the differences in k'f2i between RTN4 and AMR were
statistically significant in the case ofone and three syl-
lables (p<.O1), and the differences between RTN3 and
AMR were statistically significant only in the case of
one syllable Cp<.05). The differences in -.z between
JPN and R[[N3 were statistically significant when ISII
was two syllables and three syllables (p<.O1), The difL
-3e-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety ofJapan
A "Critical
Period" in Second Language Acquisition Reconsidered
[[lable6 Percentages of shortening for the target stressed vowels (vowel shortening from a one-syllable ISI to a two-
syllable ISI and three-syllable ISI).
ISIsBetween words Within a word
1.2 1-3 1.2 1.3Mean
AMRJPNRTNRTNIRTN2RTN3RTN41741715221717 28132826312927 26le2320292521 33223636363834 26122624302725
Notc. 1-2==a one-syllable ISI to a two-syllable ISI; 1-3=a one-syLlable ISI to a three-syllable ISI.
ferences in f,9I between JPN and RTN4 were statisti-
cally significant when ISI1 was two and three syllables
(p<,Ol). Regarding the difference in kSJ'2i betweenRTN3 and RTN4, the statistical differenee was only
fbund when ISI1 was three syllables (p<.Ol). In addition to the chirational patterns related to each
ef the subject groups in terms ofvalues of -fi,fl-i
, we ob-
tained the average increment of increase in the norrnal-ized ISI durations from one-syllable ISIs to two-sylla-ble ISIs. The average increment in the normalized ISIduration per syllable, as estimated from the correlationcurve, shows that normalized ISI durations increasedby a larger increment fbr JPN, rather than RTN and
AMR. The results of analysis of variance showed sig-
nificant differences in the subject groups (ISIs within a
word: F(2, 140)= 8.88, p<.05, ny2=,11; ISIs betweenwords: iF<2, 143)=16.06, p<,Ol, n2=,18). The results
of multiple comparison tests (TUkey-Kramer) showed
sigriificant differences between JPN and RTN (ISIswithin a word: p<.05; ISIs between words: p<.Ol).Regarding RTNI, RTN2, JPN, and AMR, analysis of
variance showed statistieally significant differences
among the subject groups in the increment of increase
both in ISIs within a word (fl[3, 139)=8,78, p<,Ol,q2=,16) and ISIs between words (F<3, 142)=13,14,
p<,Ol, rp2=.22), The results ef multiple comparison
tests (Tukey-Kramer) showed significant differencesbetween JPN and RTNI Cp<.OI). Also, analysis of
variance shQwed that there were significant differences
among AMR, JPN, RTN3, and I(rN4 (ISIs within a
word: F<3, 139)=5,88,p<.Ol, n2=,11; ISIs betweenwords: F(3, 142)=10.82,p<.Ol, op2=.19). RTN3 and
JPN showed the statistically significant differences in
the increment ofincrease both in ISIs within a word and
ISIs between words (p)<,Ol). RTN4 and JPN showed
statistically significant differences in the increment ofincrease both in ISIs within a word (p<,Ol) and ISIsbetween words ip<,05), Turning to the target vowel shortening, Table 6shows the degree of vowel shortening fbr AMR, JPN,and RTN. That is, it shews the percentages of shorten-
ing fbr the target stressed yowels from a one-syllableISI to a two-syllable ISI and three-syllable ISI. Asshown in the table, the stressed vowels were shortened
considerably when unstressed syllables were added
to the ISIs in the case of AMR and RTN, while JPNexhibited a much smaller percentage of shortening. Theaverage percentages ofvowel shortening were 26% for
AMR, 2696 fbr RTN, and 129i6 for JPN. The percentage
of the target vowel shortening was shown to be higherin the case ofISIs within a word, as compared to ISIsbetween words fbr all subject groups, The results of
analysis of variance showed significant differences inthe percentage of shortening among AMR, JPN, andRTN (ISIs within a word: one-syllable ISIs to two-
syllable ISIs, F(2, 47)=623,p<.Ol, rp2 =,21; one-syl-
lable lSIs te three-syllable lSIs, F<2, 47)=7,68,p<.Ol,
rp2=,25; ISIs between words: one-syllable ISIs to two-
syllable ISIs, F(2, 47)=]1.28, p<.Ol, u2=.32; one-
syllable ISIs to a three-syllable ISIs, F(2, 47)=7.33,
p<.05, ny2=.24). The results of multiple eomparison
tests (Tukey-Kramer) showed that the differences in
the percentage of shortening between AMR and JPNwere statistically significant in all cases (p<,Ol), ex-
cept that the vowels from a one-syllable ISI to a three-syllable ISI between words were statistically significant
-31-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
fiI2retN (ResearchArticles)
40%uaa:S
30%tsets
20%&sg
lo%ee
o%
Number ofsyl]ables in ISII
Figure 4 Percentage of shortening of stressed veyvels in
the ISIs between words forAMR, JPN, RTNI,
and RTN2.
Cp<.05), The differences between RTN and JPN were
statistically sigriificant in all cases (p<.O1), except for
the vowels from a one-syllable ISI to a two-syllable ISI
within a word being significant (p<.05), while thosebetween AMR and RTN were not significant.
Rega:;ding the effbct of the age of arrival, we com-
pared the target vowel shortening for RTN1 and RTN2.Figure 4 shows the percentage of shortening of stressed
vowels in the ISIs between words fbr AMR, JPN,RTNl, and RTN2. As shown in the figure, we observe
that the subgroups of RTN-RTNI and RTN2-and
AMR showed a similar tendency in the percentage efthe target vowel shortening. The average percentage ofvowel shortening from a one-syllable ISI to a two-syl-
lable ISI was 18% for RTNI (ISIs within a word: 20%,
ISIs between words: 15%). In the case of RTN2, the
average percentage ofvowel shortening was 26% (ISIswithin a word: 29%, ISIs between words: 22%). Theresults of analysis of variance showed significant dififerences among the subject groups (ISIs within a word:
one-syllable ISIs to two-syllable ISIs, F(3, 46)==5,13,
p<.Ol, op2= 25; one-syllable lSIs to three-syllable lSIs,fJ<3, 46)=4,61,p<,05, n2=.23; ISIs between words:one-syllable ISIs to two-syllable ISIs, F<3, 46)=8.97,
p<.Ol, ny2=.37; one-syllable lSIs tQ three-syllable lSIs,F<3, 46)=4.96,p<.05, op2=.24). The results ofmultiple
comparison tests (Tukey-Kramer) showed that the
differences between AMR and RTNI, those betweenAMR and RTN2, and those between RTNI and RTN2were not statistically significant. The difference in the
percentage of vowel shortening between RTNI and
JPN was statistically significant in the one-syllable ISIs
to two-syllable ISIs between words (p<.O1), and those
in the one-syllable ISIs to three-syllable ISIs within a
werd and between words (p<,05), The one-syllable
40%uag.-S
30%tsSts
20%&sg
lo%e
£ o%
Number ofsy}lables in ISII
Figure 5 Percentage ofshortening ofstressed vowels in
the ISIs between words for AMR, JPN, RTN3,
alld Fa]N4.
ISIs to two-syllable JSIs within a word did not show
statistical differences between RTNI and JPN, Also,
the differences in the percentage ef vowel shortening
between RTN2 and JPN were statistically significant
(one-syllable ISIs to two-syllable:p<,O1; one-syllable
ISIs to three-syllable ISIs:p<.05).
Regarding the effect of the length of residence, we
compared the target vowel shortening for RTN3 and
RTN4. Figure 5 shows the percentage of shorten-
ing of stressed vowels in the ISIs between words fbr
AMR, JPN, RI'N3, and IU'N4, As shown in the figure,we observe that the subgroups of RTN-RTN3 and
RTN4and AMR showed a similar tendency in the
percentage ofthe target vowel shortening. The average
percentage ofvewel shortening from a ene-syllable ISI
to a two-syllable ISI was 21% for RTN3 (ISIs within a
word: 25%, ISIs between words: 17%). In the case of
RTN4, the average pereentage ofvowel shortening was
19% (ISIs within a word: 219t5, ISIs between words:17%). The results of analysis of variance showed
significant differences among the subject groups invowel shortening (ISIs within a word: ene-syllable ISIsto two-syllable ISIs, F(3, 46)=4.15, p<,05, ag2;21;one-syllable ISIs to three-syllable ISIs, R3, 46) ==4,80,
p<.Ol, ny2=.24; ISIs between words: one-syllable ISIsto two-syllable ISIs, F(3, 46)=7.23, p<.Ol, op2=.32;one-syllable ISIs to three-syllable ISIs, F<3, 46)= 4,66,
p<,05, op2=.23). The results of multiple comparison
tests (Tukey-Kramer) showed that the differencesbetween AMR and RTN3, those between AMR and
RTTN4, and those between RTN3 and RTN4 were not
statistically significant. The difflerences in the percent-age ofvowel shortening between RTN3 and JPN were
statistically significant in all cases (pJ<,Ol), except inone-syllable ISIs to two-syllable ISIs within a word and
-32-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
A "Critical
Period" in Second Language Acquisition Reconsidered
one-syllable ISIs to three-syllable ISIs between words
ip<.05). Also, the differences in the percentage of
vowel shortening between IU]N4 and JPN were statisti-
cally significant in all but one case ip<,05 in all cases
except in one-syllable ISIs to twe-syllable ISIs betweenwerds:p<,Ol).
4. Discussion
In the present study, we examined the L2 productionof Japanese learners in terms of ISI durational control
and stressed vowel shortening, In both parameters of
prociuction, we observed significant differences be-tween the JPN wnhout the experience of living in thetarget language county and RTN with the experience.
The experience ofliving in the target language countrywas shown to be a crucial factor in the acquisitiQn of
production patterns. The quantity and quality of L2input are different between JPN and RTN. Regardingthe quantity of input, Asher (1977) reported that on
average, children, by age 6, haye spent a minimum
of 17, 520 hours (8 hours a day) listening to their L1,while Japanese L2 learners, by age 18, have spent from1,OOO to 2,OOO hours (20 minutes a day) of contact
with English as L2 (Thkefuta 1997). In addition to
significant differences in quantity of input, it is obvi-ous that the quality of input is different when learn-ers are surreunded by native speakers of the target
language, in contrast to when they are not exposed to
native speakers in everyday life. In the present study,the subjects of RTN had lived in the U.S. for over ene
year at least. This suggests that fbr successfu1 acquisi-
tion, a minimum of one-year living might be requiredas the amount of necessary exposure to the target
language, which is in accordance with the results of
our study with junior high school students (Sudo andKaneko 2005). Several studies on short-term studying
abroad programs haye reported positive effects on the
participants such as global awareness, socio-cultural
adaptation, and academic success, but not immediateeffects on speaking and listening abilities ofthe partici-
pants (Antonakopoulou 2013, Kurt, Olitsky and Geis
2013, Xu, DeSilva, Neufeldt and Dane 2013), lending
support to the necessity ofa minimum longer period forthe improvement ofthose abilities.
Regarding an early-late efTect on L2 production,we observed that RTNI, rather than RTN2, showed
a significant amount of approxirnation to the produc-tion pattems ef native speakers. We found significantdifferences in the ISI durational patterns between thesubgroups of RTN which were elassified according to
the age of arrival, RTNI being the subgroup of RTN
whose age of arrival was befbre puberty and RTN2, the
subgroup whese age of anival was after puberty. Thissignificant early-late age effect in the ISI production isin line with the results of the previous studies (Bakerand [I)rofimovich 2005, Darcy and Kruger 2012, Flegeet al, 1999, Guion 2003, McCanhy et al. 2013) whichreported the early-late effect on production. Theseresults confirm that there is a critieal period for the ac-quisition ofthis specific feature ofproduction.
Turning to another parameter ofproduction, stressed
vowel shortening, we found that an early-late age efi
fect was not significant on this parameter ofproduction
patterns. Concretely, we did not observe a significant
difference among the subgroups of RTN, RTNI, 2, 3,and 4, The two factors of age of arrival and Iength ofresidence in the target language country did not result
in significant differences in the production patterns ofstressed vowel shortening. This result suggests that
this specific feature ofproduction can be acquired after
the critical period, supponing the results ofthe studies
(Bongaerts et aL 1995, Neufeld 1977, 1979, 1980),which reported even late learners can acquire a native-
like productien. We have presented a difference in age effect on thetwo production parameters. The results indicate that
the degree of diMculty in acquiring durational control
of stressed vowels is not so high, compared with the
acquisition of the ISI durational control, This suggeststhat the acquisition of unstressed syllables in an ISIis more ehallenging than that ef stressed syllables forJapanese learners of English, ofiering further supportto the results ofour study reporting difficulty in acquir-ing durational control ofunstressed syllables (Sudo andKaneko2005).
In addition to production parameters, we also mea-
sured the English proficiency of Japanese learners bythe TOEIC scores. We observed significant differencesin the total TOEIC scores and those in both the Listen-ing Section and the Reading Section between JPN and
RTN. The results of the TOEIC scQres showed that the
experience ofliving in the target language country was
a factor which can facilitate the acquisition oflistening
and reading proficiencies. Regarding the subgroups
of RTN, which were classified accerding to the ageof arrival, we observed significant differences in theTOEIC scores between the twD subgroups. Concretely,a significant difference between RTN1 and RTN2 was
observed in the Listening Section, but not in the Read-ing Section. The results indicated that the factor of age
can be complemented by other factors of learners in
-33-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety of Japan
bcrfithSc(ResearchArticles)
reading ability, while the age factor is a decisive factorin the case of listening ability. This result of the early-
late age effect on listening ability is in accordance with
the results efprevious studies (Flege et al. 1999),
In this study, we attempted to find out why previous
studies on an age effect reportecl different results about
an early-late factor. We offbred data showing that L2Japanese learners have different degrees of attainment
of L2 skills, depending on which feature of productionis the target of acquisition. Stressed vowel shortening
was shown to be a feature which can be acquired after
puberty, while ISI durational control, especially fbr theunstressed syllables in the ISI, showed an effect of the
early-late dichotomy. Regarding listening and reading
abilities measured by the TOEIC, listening proficiencywas shown to have an eaTly-late age efTbct, while read-
ing proficiency could be acquired after puberty. To sum
up, the results of this study lend partial support to the
existence ofa "critical
period" in L2 acquisition.
The factor of length ofresidence in the target lan-
guage country was shown to have an effect on ISI
durational control, but not on compensatory vowel
shortening, We observed that RTN3 with the length
of residence of six years or more, rather than RTN4,showed a significant arnount of approximation to the
production patterns of native speakers, The factor oflength of residence was shown to be not signifieant
in the TOEIC soores of RTN3 and RTN4, which is
in accordance with the results of r[lahta
et al. (19gl),revealing that length of residence was not identified
as a significant predictor. In other words, we observed
significant difTerences in the TOEIC scores of the
Listening Section between IU]Nl and RTN2, but notbetween RTN3 and RTN4, Although the number of
the subjects in this study was net large, and the ageof arrival and the length of residenee were not quan-titatively matched with each other, the results can still
be interpreted as indicating that the age of anival is a
factor that overrides the length ofresidence in regard to
the English proficiency measured by the TOEIC scores.
Even if the length of residence was not long, some of
the leamers showed proficient listening and reading
abilities. It should be noted that this factor could be
complemented by other factors ofIearners, such as mo-
tivation, aptitude, attitude, personality among others,
provided that the learners have the experience of living
in a target language country, regardless of the length
of time. Here, we should consider some limitations of
this study. There remains a possibility that the different
degrees of significance observed in the age of arrival
and the length of residenee may be attributed to the
combined effects of the twe factors. Further research,
with a larger number of subjects, is needed, especially
to analyze the main factor in different degrees of acqui-
sition attainment.
5. Conclusien
The present study investigated the hypothesis of
an age effect, the Critical Period Hypothesis, on L2
production and listening!reading abilities by Japanese
learners of English, and showed that a critical periodexists fbr some aspects of L2 acquisition, but not fbr
others. Concretely, we fbund a critical period for the
acquisition of listening ability and ISI durational con-
trol, including production of unstressed syllables; we
also found that specific features ofproduction, such as
stressed vowel shortening and reading ability could be
acquired after the critical period.
Acknowledgement
This research was supported by a Grant-in-Aid for
Scientific Research from the Ministry of Education,
Culture, Sperts, Science and Tlechnolegy [#25370700].
References
Altman, H. (1980) tLForeign
language teaching: Focus on the
learner.7' In H. Altman and J. Vaughan (eds.) Fbreign
language teaching: Mlaeting individual needs, 1-16. 0x-
ford: Pergamon.
Antonakopoulou, E. (2013) "Sociocultural
adaptation ofU. S.
education abroad students in Greece: The efTects ofpro-
gram duration and intervention," F>'ontiers 23, 60-72.
Asher, J. J. (1977) "The
strategy of the total physical re-
sponse: An application to learning Russian." 77ie Mbct-
ernLanguageJburnai44(3),291-300,
Asher, J, J. and R. Garcia (1969) `"The
optimal age to learn
a foreign language." 7ke Mbdern Language Jbwrnal
53(S),334341,
Baker, W, and P, Trofimovich (2005) ttlnteraction
of native-
and second-language vowel system(s) in early and late
bilinguals,]' Language and EPeeeh 48(1), 1-27.Beckman, M. E, (1992)
`tEvidcnce fbr speech rhythms across
languages." In Y. Tohkura, E, Vatikiotis-Bateson and
Y, Sagjsaka (eds.) EPeech perceptton, produetion and
linguisticstnetcture,457463.Tokyo:Ohmsha.Best, C, T,, G. W. McRoberts and E, Goodeli (2001)
"Dis-
crimination ef the non-natlve consonant contrasts vary-
ing in perceptual assimilation to the listener's native
phonological system." 71Te Jburnal of the Acoustical Society ofAmeriea 109(2), 775-794.
Bongaerts, T., B. Planken and E. Schils (1995) "Can
late
-34-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety ofJapan
A"CriticalPeriodi:in Second Language Acquisition Reconsidered
starters attain a native accent in a foreign language? A
test of the critical period hypothesis." In D. Singleton
and Z. Lengycl (cds,) 71ie agefactor in second languqge
acguisition, 30-SO. Cleyedon: Multilingual Matters.
Bomstein, M, H. (1989) "Sensitive
periods in development:
Stmctural characteristics and causal interpretations."
RsychologicalButletin 10S(2), 179-197.
Bozorgian, H, (2012) "The
relationship between listening and
other language ski1}s in International English Language
Testing System." 7:heor]; and Practice in Language
Studies 2(4), 657-663.
Darcy, I, and F, Kruger (2012) "Vowel
perception and pro-
duction in Turkish children acquiring L2 German." Jburnat ofPhonetics 40(4), 568-S8I .Dauer, R. M. (1983)
`CStress-timing
and syllable-timing re-
analyzed," Jburnal oj'Phonetics l l(1), 5 1-62.
DeKeyser, R. (2000) "
[he robustness of critical period efi
fects in second language acquisition." Simdies in Sbcond
LangwageAequisition22,499-533.
Ellis, R. (1994) 71!e stucly ofsecond language acquisition.
Oxford: Oxford University Press.
Ericksson, A. (1991) Aspects of Swedish speeeh rllythm.
Gothenburg Monographs in Linguistics 9, University of
Goteborg.Fant, G,, A, Kruckenberg and L. Nord (1991)
L`Durational
correlates of stress in Swedish, French, and English.'i
Jburnal ofPhonetics 19, 351-365.Flege, J. E. (1995)
"Second-language
speech learning: Theo-
ry, findings, and problems." In W. Strange (ed.) SPeech perception and liprguistic experience: lssues in cross-
linguistic researeh, 233-277. Timonium, MD: York Press,
Flege, J, E, (1999) "Age
of leaming and second-language
speech," In D, Birdsong (ed,) Second language aceui-
sition and the Critieal Period H)lpotkesis research,
101-1 32. HMsdale, NJ: Lawrence Er!baum.
Flege, J. E. (2002) "Interactions
between the native and
second-language phonetie systems." In P. Bunneister,
T. Piske and A. Rohde (eds.) An integrated view of lan- guage devetopment: Papers in honor ofHemming Pft)de, 217-244. Trier: Wissenschaftlicher Verlag Trier,
Flege, J. E. and S. Liu (20el) "`The effect of experience on
adults' acquisition of a second language." Studies in
SecendLangvageAcqwisition23,527-S52,Flege, J. E,, I. R, A, MacKay and D. Meador (1999)
"Native
Italian speakers' perception and production of English
vowels." 71lfe Jburnal of the Acoustical Society ofAmer- ica 106(5), 2973-2987,
Flege, J. E., M. J. Muuro and I. R. A. MacKay (199S) "The
effect of age ofsecond language Iearning on the procluc-
tion of English consenants." SPeech Communication 16(1), 1-26,
Flege, J. E., C. Schirru and I. R. A. MacKay (2003) "Interac-
tion between the native and second language phonetic subsystem." SPeech Commornication 40(4), 467-491.Fowler, C, A. (1977) avming controt in speech production
(Unpublished doctoral dissertatien), University of Con-
necticut, Storrs, CT.Grabe, E. and E. L. Low (2002)
"Durational
variability in
speech and the Rhythm Class Hypothesis," Papers in
Laboratoty Phenology 7, S 1 5-546,
Guion, S. G. (2ee3) "The
vowel systems ef Quicha-Spanish Bilinguals." Phonetica 60(2), 8-128.
Huggins, A. F. W. (1975) "On
isochrony and syntax," In
G. Fant and M. A. A, Tatham (eds.) Auditor), analysis
andperception ofspeech, 45S-464. London: Academic
Press.In'nami, Y, and R, Koizumi (2012)
"Factor
structure of the
revised TOEICg test: A multiple-sample analysis." Lan-
guage Tlasting 29(1), 131-152.
Katz, J. (2012) `CCompression
effects in English." .Iburnal of Phonetics40(3),390-402,Kinsella, C. (2009) An investigation into the profcienqy of successul late leamens qf'fi}'ench (Unpub1ished doctoral dissertation). Trinity College Dublin.
Knudsen, E. L. (2004) "Sensitive
periods in the development
ofthe brain and behavior." Jburnal of Cognitive Nleuro- science 16(8), 1412-1425,
Kurt, M., N. Olitsky and P. Geis (2013) `'Assessing
global awareness over short-term study abroad sequence: A
factor allalysis," F)'ontiers 23, 22-4 1 .Lamendella, J. (1977)
"Generai
princip!es ofneurofunctional
organization and their manifestatiolls in primary and
non-primary language acquisition." Language Learning
27(1), 15S-196,
Larsen-Freeman, D. and M. H. Long (1991) An introduction
to second language acquisition research. London: Long-
man,
Laver, J, (1994) Principles ofphonetics. Cambridge: Cam-
bridgeUniversityPress,Lehiste, I. (1977)
"Isochrony
reconsidered," Jburnal qfPho- netics S(3), 2S3-263.Lenneberg, E. (1967) Biological founclations of langttage. New York, NY: Wiley.Leng, M. (1990)
"Maturational
constraints on language de-
velepment." Studies in Second Langitczge Aequisition
12(03),251-285,
Majer, R, C, (2001) 1 breign accent 7he ontogeay andpdy-
legeay of second language phonology. Mahwah, NJ: Lawrence Erlbaum Associates,McCarthy, K, M,, B. G, Evans and M, Mahen (2013)
"Ac-
quiring a secend language in an immigrant community;
The production of Sylheti and English stops and vowels
by London-Bengali speakers." .lburnal of Phonetics 41(5), 344358.
Mochizuki, A. A., M. M. Sudo, K. Itoh and E. Kirino (2013)
`LSystem
analysis ofphysical abilities, secial functioning
and language proficiency." Jburnal oj'llbalth and SPorts Science Jdentendo 5(1), 1-9.
Mochizuki-Sudo, M, and S, Kiritani (1991) "Production
and perception of stress-rclated durational patterns in
Japanese learners of English." jburnal ofPhonetics 19,
231-248.Munhall, K,, C, Fowler, S, Hawkins ancl E. Saltzman (1992)
-35-
Phonetic Society of Japan
NII-Electronic Library Service
PhoneticSociety ofJapan
blxff.!(ResearchArticles)
"`Compensatory
shortening' in monosyllables ofspoken
English." Jburnal ofPhonetics 20, 225-239.
Neufeld, G. G. (1977) "Language
learning ability in adults:
A study on the acquisition of prosodic and articulatory
features." PPlorking Popers on Bilingualism 12, 45-60.Neufeld, G. G. (1979)
"Toward a theory ef language Iearning
ability." Language Learning 29(2), 227-24 I .Neufeld, G. G. (1980) L`On
the adu]tis ability to acquire pho- nelogy." 71ESaL 2uarterly 14(3), 285-298.
Oyama, S. (1976) `CA
sensitive period fbr the acquisition efa
nonnative phonological systern." Jburnai oj' Rsp,cholin-
guisticResearch5C3),261-283.Penfield, W. and L. Roberts (1959) speech and brain-mecha-
nisms. Princeton, NJ: Princeton University Press.
Powers, D, E, (201O) `:Valjdity:
What does it mean for the
TOEIC# tests?" 7Z)lllC Conrpendium. Princeton, NJ: ETS.Rakerd, B. W, Sennett and C. A. Fowler (1987)
"Domain-
final lengthening and foot-level shortening in spoken
English." Phonetica 44(3), 147-]55.
Scovel, T. (1981) "The
recognitien of fbreign accents in Eng-
lish and its implications for psycholinguistic theeries of
language acquisitien," In J,-G. Savard and L. Laforge
(eds,) Proceedings of the 5th international Association
of Appiied Linguistics, 389-401. Montreal, QC: Laval UniversityPress.Shaiman, S. (2001)
"Kinematics of compensatory shortening:
The eflect of speaking rate and coda conditien on intra-
and inter-articulatory timing." Jburnal of Phonetics 29(1), 89-107.Singleton, D. (2005)
`LThe
critical period hypethesis: A coat
of many colours." Ihternationat Review ofApptied Lin-
guistics 1O(4), 269-28S.Skehan, P. (1989) Individual di'nerences in second-tanguage
learning. London: Edward Arnold.
Snew, C. E. (1987) "Relevance
of the notion ofa critical pe-
riod to language acquisition.'i In M. H. Bornstein (ed.) Sensitive periods in deveiopment.' interdisciptinary per-
spective, 183-209. Hillsdale, NJ: Lawrence Erlbaum.Snow, C. E. and M. Hoethagel-Hohle (1977)
"Age
differ-
ences in the pronunciation of foreign sounds." Language
and speech 20(4), 357-365.
Sudo, M. M, and I, Kaneko (2005) "Acquisition process of
English rhythmic patterns: Comparison between native
speakers ofEnglish and Japanesejunior high school stu-
dents.i'MC:ETBulletin 40, 1-14.
Sudo, M. M. and I. Kaneke (2012) "Is
there a [critical
period7
in acquisition of English by Japanese learners? Factors
affecting TOEIC sceres and the production of English
rhythmic patterns." .L4 CETJburnal S4, 2 1-32.
Sudo, M. M. ancl I. Kaneko (2013) "Kaigaitaizaikaishinenrei
to kikan ga nihoajin gakushusha ni yoru eigo boin no
cornpensatory shortening ni ataeru eikyo [Effects of
age of arrival and living duration in the target 1anguage
country on compensatory vowel shortening by Japanese learners of English]," Proceedings of the 2013 SPring A4betingoftheAcoustiealSocietyof.rL-)an,451-452.
Sudo, M. M., I. Kaneko and H. Nishimura (2012) "Nihoniin
gakushusha ni yoru eigoboin no compensatory shorten-
ing no shutoku [Acquisition of compensatory vowel
shortening by Japanese leamers of English]." Proceed-
ings ofthe 2012 Autumn ildleeting qfthe Aeousticat Soci- eo,ofJopan,355-356.
Sude, M. M., L Kaneko and H. Nishimura (2013) "Eigo no
seisei pattern to choukairyoku doklcairyoku ni okeru
nenrei yeuin: Nihenjin kikokusei to ippan daigakusei
no bunseki [Effects on age factor on English production patterns and listening/reading proficiency: An analysis
of Japanese retumees and college students]," Proceed-
ings ofthe 2013 Autumn Adleeting ofthe Acoustical Soci- ety ofJopan, 283-284.
Sudo, M. M., A. A. Mochizuki, K. Itoh and E. Kirino (2011)
"Effects of physical performance and language pre-
ficiency on brain activation: From a social cognition
viewpoint," inatrgural international Acacieay of SPor- tology 1, 36-40.
Sundara, M,, L. Polka and S. Baum (2006) t`Production
of
coronal stops by simultaneous bilingual adults." Bitin-
gualism.' Language and Cbgnition 9(1), 97-1 14.Tahta, S., M. Wood and K, Loewenthal (1981)
"Foreign ac-
cents: Factors relating to transfer of accent from the first language to secend language.i' Language andth)eech 24,
265-272.Takefuta, Y. (1997) Eigokyoiku no kngaku [The scientific ap-
proach to English education]. Tokyo: Kenkyusha.
Xu, M,, C. DeSilya, E. Neufeldt and J. H. Dane (2013) "The
impact of study abroad on academic success: An analy-
sis of first-time students entering Old Dominion Univer-
sity, Virginia, 2000-2004." Frontiers 23, 90-103.
(Received Sep. 5, 2014, Accepted Oct. 14, 2015)
-36-