language aptitude and long-term achievement in early childhood l2 learners
TRANSCRIPT
Applied Linguistics 2014 354 483ndash503 Oxford University Press 2014
doi101093applinamu013 Advance Access published on 23 April 2014
Language Aptitude and Long-termAchievement in Early Childhood L2Learners
GISELA GRANENA
School of Languages Universitat Oberta de Catalunya
E-mail ggranenauocedu
Language aptitude comprises cognitive and perceptual abilities that predispose
individuals to learn well or rapidly (Carroll 1981 Doughty et al 2007) and that
can interplay with age by moderating its effects on learning outcomes Although
there is agreement regarding the importance of aptitude findings have been
mixed regarding its role in child second language acquisition (eg
Abrahamsson and Hyltenstam 2008 vs DeKeyser 2000) This study set out to
investigate whether aptitude played a role in ultimate morphosyntactic attain-
ment by a group of early childhood learners Participants completed a speeded-
response and a non-speeded-response grammaticality judgement test (GJT) and
the LLAMA aptitude test (Meara 2005) Results showed the presence of an
interaction between aptitude test and target structure Aptitude was signifi-
cantly related to early learnersrsquo attainment in structures involving grammatical
agreement on the non-speeded-response GJT This was interpreted as showing
the compensatory role of aptitude in structures for which age effects are the
strongest and a relationship between measures of aptitude weighted in favour of
explicit cognitive processes and untimed measures that induce learners to
approach language analytically
BACKGROUND TO THE STUDY
Long-term achievement in second language (L2) acquisition is characterized
by high inter-individual variability Such variability in learning outcomes has
been attributed apart from age to a broad array of external socialenviron-
mental factors and to individual learner differences in the cognitive and affect-
ive domains In naturalistic contexts where acquisition takes place through
immersion in the L2-speaking environment several of these learner and en-
vironmental variables have been investigated for their potential in moderating
the effects of age on long-term learning outcomes One of these variables has
been language aptitude a complex construct that comprises cognitive and
perceptual abilities that predispose individuals to learn well or rapidly
(Carroll 1981 Doughty et al 2007) Its relevance to second language acquisi-
tion (SLA) has been consistently shown by an increasingly larger number of
studies that have recently made aptitude research a revitalized area of study
spurred by advances in cognitive and educational psychology in
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
understanding how languages are learned as well as by recent developments
in the measurement and conceptualization of aptitude (eg Doughty et al
2010 Granena 2013a)
Skehan (1989) in fact argued that aptitude could be even more relevant for
learning in naturalistic (acquisition-rich) contexts than in instructed contexts
because of the greater amount of input that the learner has to process and the
pressure to discover regularities and make generalizations merely from L2 ex-
posure The nature of such a relationship however is still controversial and
research findings have been mixed Some studies have found that aptitude is
differentially related to learning outcomes in early and late L2 learners
Specifically these studies found that aptitude was related to variation in late
learnersrsquo morphosyntactic attainment but not in early learnersrsquo attainment
(DeKeyser 2000 DeKeyser et al 2010) or that different components of aptitude
were not equally related to L2 attainment in the two populations (Harley and
Hart 1997) All these studies have provided evidence in support of an interaction
between age and language aptitude Other studies however have found no
evidence of such an interaction either because aptitude played a similar role in
both early and late learnersrsquo morphosyntactic attainment (eg Abrahamsson
and Hyltenstam 2008) or because it did not play a role in either early or late
learnersrsquo morphosyntactic attainment (Granena and Long 2013)
The study by Harley and Hart (1997) is commonly cited as the first study that
documented an interaction between age and aptitude in early and late L2
learners The study conducted with learners in immersion programmes in
Canada found significant positive correlations between memory and L2
outcomes in early immersion learners and between analytical ability and
L2 outcomes in late immersion learners Although this study is usually cited
as evidence in support of an interaction between age and aptitude the fact
that early and late learners were exposed to two different types of instruction
(ie holistic memory-based vs language analysis) allows for an alternative
competing explanation since the relationship between age and aptitude
could be in fact the result of an interaction between aptitude and instructional
method
Unlike Harley and Hart (1997) the two studies by DeKeyser (DeKeyser 2000
and DeKeyser et al 2010) were carried out with long-term residents in an
L2-speaking country In DeKeyser (2000) Hungarian speakers of L2 English
in the USA were tested on various elements of morphosyntax via an auditory
grammaticality judgement test (GJT) Aptitude was operationalized as L1
verbal analytical ability The results showed a significant correlation between
GJT scores and language aptitude among late arrivals (ie participants who had
arrived in the country after age 15 years) but a non-significant correlation
among early arrivals Those participants who were late arrivals and scored
within the range of child arrivals or came close were all high-aptitude par-
ticipants The exception was a participant who did not have high aptitude but
who nevertheless scored within the range of child arrivals on the GJT
DeKeyser argued that this participant a postdoctoral student in the natural
484 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
sciences was probably of above-average analytic ability but that his true skills
had not been captured by the aptitude measure used the Words-in-Sentences
subtest in the Modern Language Aptitude Test (MLAT Carroll and Sapon
1959) On the basis of these results DeKeyser concluded that above-average
analytic abilities are required to reach near-native levels in the L2
DeKeyser et al (2010) replicated the findings in DeKeyser (2000) with
Russian speakers of L2 English and L2 Hebrew in the USA and Israel respect-
ively In these two parallel cross-linguistic studies aptitude [operationalized as
L1 verbal aptitude and measured by a test comparable with the verbal
Scholastic Aptitude Test (SAT)] was significantly correlated with ultimate at-
tainment for the adult learners but not for the early learners Specifically the
significant correlation was found for the 18ndash40 years age of acquisition range
but not for the age of acquisition lt18 years group The correlation in the age of
acquisition gt40 years group was not significant either which was interpreted
as the result of factors related to aging
While language aptitude was not related to ultimate morphosyntactic attain-
ment among early acquirers in the studies by DeKeyser (2000) and DeKeyser
et al (2010) Abrahamsson and Hyltenstam (2008) and Bylund et al (2012)
found aptitude effects in pre-pubescent L2 learnersrsquo ultimate attainment who
were first exposed to the L2 before age 12 years In Abrahamsson and
Hyltenstam (2008) there was a significant and moderately strong relationship
between aptitude and ultimate attainment in the early L2-learner group
In this group 72 per cent of the learners who performed within the native
speaker (NS) range had high aptitude operationalized as being above
the grouprsquos average level of aptitude In the group of late learners the four
learners who were able to score within the NS range were all also above
average in terms of aptitude In Bylund et al (2012) aptitude was reported
as the only significant predictor for nativelike performance in both the L1
and the L2
Like DeKeyser (2000) and DeKeyser et al (2010) Abrahamsson and
Hyltenstam (2008) and Bylund et al (2012) relied on grammaticality judge-
ments to measure ultimate attainment Abrahamsson and Hyltenstam (2008)
combined the scores of two different GJT modalities auditory and written
while Bylund et al (2012) used an untimed cloze test and an auditory GJT
that gave participants a time frame of 10 s to respond after the end of each
sentence Unlike DeKeyser (2000) and DeKeyser et al (2010) where aptitude
was operationalized as L1 verbal ability Abrahamsson and Hyltenstam and
Bylund et al used a non-L1-based omnibus test of aptitude the Swansea
Language Aptitude Tests (LAT Meara et al 2003) a combination of five dif-
ferent subtests phonetic memory analytical ability grammatical inferencing
sound recognition and soundndashsymbol correspondence On the basis of the
results of their study Abrahamsson and Hyltenstam (2008) concluded that
lsquoDeKeyserrsquos claim that language aptitude lsquolsquoonly plays a role for adult learnersrsquorsquo
should be modified to state that language aptitude plays not only a crucial role
for adult learners but also a certain role for child learnersrsquo (p 499)
G GRANENA 485
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
However another study Granena and Long (2013) did not find an effect of
aptitude in either early or late learnersrsquo morphosyntactic attainment Like
Abrahamsson and Hyltenstam (2008) they used a non-L1-based omnibus
measure of aptitude the latest version of the LAT the LLAMA aptitude test
(Meara 2005) but unlike Abrahamsson and Hyltenstam (2008) DeKeyser
(2000) and DeKeyser et al (2010) they measured morphosyntactic attain-
ment by means of a battery of five different tests which were used to create
an overall composite score This battery included a GJT but also an oral nar-
ration task two word order tasks and a gender assignment task Granena and
Long (2013) concluded that the conflicting results of the studies were mostly
due to methodological differences concerning sampling and sample sizes as
well as the aptitude tests procedures and outcome measures employed
Aptitude as measured by tests such as the MLAT or the LLAMA loosely
based on the MLAT may be important in morphosyntactic attainment when
this is assessed by means of tasks that focus on language forms and language
correctness such as GJTs However findings may be different when L2 attain-
ment is measured by means of oral production tasks or other meaning-based
tasks that do not call for the same analytic andor metalinguistic abilities that
also characterize explicit language aptitude measures
The issue of whether a cognitive factor such as language aptitude plays a role
in early L2 acquisition has relevant implications for SLA theory and practice
The rationale behind DeKeyserrsquos (2000) claim in support of a differential role
of aptitude in child and adult L2 acquisition was Bley-Vromanrsquos (1988 1990)
Fundamental Difference Hypothesis according to which there is a qualitative
difference between the learning mechanisms of child and adult L2 learners
younger learners learn mostly implicitly using domain-specific mechanisms
whereas older learners learn mostly explicitly using problem-solving or
domain-general mechanisms and therefore have to rely more on language
aptitude As a result of this qualitative difference between child and adult L2
learning DeKeyser predicted that aptitude would have a differential role in the
two populations If instead language aptitude is found to be equally related to
ultimate attainment in both child and adult L2 learners this could suggest that
child and adult L2 learners are more similar than predicted so far in terms of L2
learning processes
AIMS OF THE PRESENT STUDY
This study focused on the interplay of aptitude and age by investigating the
role of aptitude in early childhood morphosyntactic acquisition in a naturalistic
learning context Unfortunately there is no consensus in the literature regard-
ing a cut-off point between early and late L2 learners and different theoretic-
ally andor empirically motivated cut-offs have been suggested age 12 years
(eg Abrahamsson and Hyltenstam 2008) age 15 years (Johnson and Newport
1989 DeKeyser 2000) and age 18 years (eg DeKeyser et al 2010) Unlike
previous research the present study set out to investigate the role of aptitude
486 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
in early childhood acquisition exclusively defined as L2 learning that started in
the age range between 3 and 6 years Age 3 years was considered a reasonable
choice for dividing simultaneous (2L1) and sequential bilinguals (L2) based on
Meisel (2001 2009) Successive or sequential bilinguals learn one language
first in a given context and the L2 later and in a different context whereas
simultaneous bilinguals learn the two languages since birth and in the same
context According to Meisel (2009) successive language acquisition in early
childhood (as early as age 3 years) can be affected by early maturational
changes which justifies the classification of these sequential bilinguals as
child L2 learners
While there is considerable research on early L2 learners there is not much
research on early childhood learners whose L2 onset began after age 3 years
but before age 6 years From the point of view of learning mechanisms this is a
relevant L2 learner population since they are not expected to be fundamen-
tally different in terms of learning mechanisms from simultaneous bilinguals or
NSs Paradis (2009) for example discusses age 6 years in terms of changes in
explicitndashimplicit learning mechanisms since metalinguistic abilities start de-
veloping around that age Although early childhood L2 learners are not ex-
pected to be fundamentally different in terms of learning mechanisms from
simultaneous bilinguals or NSs they may crucially differ in linguistic attain-
ment and degree of variability (Meisel 2009) In addition early childhood
learners can be expected to be a more homogeneous population with respect
to variables such as type of L2 input available (eg interactional support and
scaffolding patterns) and other social factors that could play a role in
acquisition
The above considerations led to the following research questions
1 Is language aptitude related to early childhood L2 learnersrsquo ultimate mor-
phosyntactic attainment
2 If so is language aptitude equally related to ultimate morphosyntactic
attainment as measured by different language tests andor structures
METHODOLOGY
Participants
A total of 50 Chinese L1ndashSpanish L2 sequential bilinguals (42 per cent males
and 58 per cent females) participated in the study They were all long-term
residents in Madrid (Spain) with no less than a high-school diploma (80 per
cent of them had or were studying a college degree) who arrived in the coun-
try in early childhood (ie between ages 3 and 6 years) or were born in Spain1
In either case these early L2 learners had been born to Chinese-speaking
parents who did not speak the L2 at home as the participants themselves
reported in the biographical questionnaire (see Supplementary Appendix A)
As a result even those early L2 learners who had been born in Spain were not
G GRANENA 487
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
understanding how languages are learned as well as by recent developments
in the measurement and conceptualization of aptitude (eg Doughty et al
2010 Granena 2013a)
Skehan (1989) in fact argued that aptitude could be even more relevant for
learning in naturalistic (acquisition-rich) contexts than in instructed contexts
because of the greater amount of input that the learner has to process and the
pressure to discover regularities and make generalizations merely from L2 ex-
posure The nature of such a relationship however is still controversial and
research findings have been mixed Some studies have found that aptitude is
differentially related to learning outcomes in early and late L2 learners
Specifically these studies found that aptitude was related to variation in late
learnersrsquo morphosyntactic attainment but not in early learnersrsquo attainment
(DeKeyser 2000 DeKeyser et al 2010) or that different components of aptitude
were not equally related to L2 attainment in the two populations (Harley and
Hart 1997) All these studies have provided evidence in support of an interaction
between age and language aptitude Other studies however have found no
evidence of such an interaction either because aptitude played a similar role in
both early and late learnersrsquo morphosyntactic attainment (eg Abrahamsson
and Hyltenstam 2008) or because it did not play a role in either early or late
learnersrsquo morphosyntactic attainment (Granena and Long 2013)
The study by Harley and Hart (1997) is commonly cited as the first study that
documented an interaction between age and aptitude in early and late L2
learners The study conducted with learners in immersion programmes in
Canada found significant positive correlations between memory and L2
outcomes in early immersion learners and between analytical ability and
L2 outcomes in late immersion learners Although this study is usually cited
as evidence in support of an interaction between age and aptitude the fact
that early and late learners were exposed to two different types of instruction
(ie holistic memory-based vs language analysis) allows for an alternative
competing explanation since the relationship between age and aptitude
could be in fact the result of an interaction between aptitude and instructional
method
Unlike Harley and Hart (1997) the two studies by DeKeyser (DeKeyser 2000
and DeKeyser et al 2010) were carried out with long-term residents in an
L2-speaking country In DeKeyser (2000) Hungarian speakers of L2 English
in the USA were tested on various elements of morphosyntax via an auditory
grammaticality judgement test (GJT) Aptitude was operationalized as L1
verbal analytical ability The results showed a significant correlation between
GJT scores and language aptitude among late arrivals (ie participants who had
arrived in the country after age 15 years) but a non-significant correlation
among early arrivals Those participants who were late arrivals and scored
within the range of child arrivals or came close were all high-aptitude par-
ticipants The exception was a participant who did not have high aptitude but
who nevertheless scored within the range of child arrivals on the GJT
DeKeyser argued that this participant a postdoctoral student in the natural
484 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
sciences was probably of above-average analytic ability but that his true skills
had not been captured by the aptitude measure used the Words-in-Sentences
subtest in the Modern Language Aptitude Test (MLAT Carroll and Sapon
1959) On the basis of these results DeKeyser concluded that above-average
analytic abilities are required to reach near-native levels in the L2
DeKeyser et al (2010) replicated the findings in DeKeyser (2000) with
Russian speakers of L2 English and L2 Hebrew in the USA and Israel respect-
ively In these two parallel cross-linguistic studies aptitude [operationalized as
L1 verbal aptitude and measured by a test comparable with the verbal
Scholastic Aptitude Test (SAT)] was significantly correlated with ultimate at-
tainment for the adult learners but not for the early learners Specifically the
significant correlation was found for the 18ndash40 years age of acquisition range
but not for the age of acquisition lt18 years group The correlation in the age of
acquisition gt40 years group was not significant either which was interpreted
as the result of factors related to aging
While language aptitude was not related to ultimate morphosyntactic attain-
ment among early acquirers in the studies by DeKeyser (2000) and DeKeyser
et al (2010) Abrahamsson and Hyltenstam (2008) and Bylund et al (2012)
found aptitude effects in pre-pubescent L2 learnersrsquo ultimate attainment who
were first exposed to the L2 before age 12 years In Abrahamsson and
Hyltenstam (2008) there was a significant and moderately strong relationship
between aptitude and ultimate attainment in the early L2-learner group
In this group 72 per cent of the learners who performed within the native
speaker (NS) range had high aptitude operationalized as being above
the grouprsquos average level of aptitude In the group of late learners the four
learners who were able to score within the NS range were all also above
average in terms of aptitude In Bylund et al (2012) aptitude was reported
as the only significant predictor for nativelike performance in both the L1
and the L2
Like DeKeyser (2000) and DeKeyser et al (2010) Abrahamsson and
Hyltenstam (2008) and Bylund et al (2012) relied on grammaticality judge-
ments to measure ultimate attainment Abrahamsson and Hyltenstam (2008)
combined the scores of two different GJT modalities auditory and written
while Bylund et al (2012) used an untimed cloze test and an auditory GJT
that gave participants a time frame of 10 s to respond after the end of each
sentence Unlike DeKeyser (2000) and DeKeyser et al (2010) where aptitude
was operationalized as L1 verbal ability Abrahamsson and Hyltenstam and
Bylund et al used a non-L1-based omnibus test of aptitude the Swansea
Language Aptitude Tests (LAT Meara et al 2003) a combination of five dif-
ferent subtests phonetic memory analytical ability grammatical inferencing
sound recognition and soundndashsymbol correspondence On the basis of the
results of their study Abrahamsson and Hyltenstam (2008) concluded that
lsquoDeKeyserrsquos claim that language aptitude lsquolsquoonly plays a role for adult learnersrsquorsquo
should be modified to state that language aptitude plays not only a crucial role
for adult learners but also a certain role for child learnersrsquo (p 499)
G GRANENA 485
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
However another study Granena and Long (2013) did not find an effect of
aptitude in either early or late learnersrsquo morphosyntactic attainment Like
Abrahamsson and Hyltenstam (2008) they used a non-L1-based omnibus
measure of aptitude the latest version of the LAT the LLAMA aptitude test
(Meara 2005) but unlike Abrahamsson and Hyltenstam (2008) DeKeyser
(2000) and DeKeyser et al (2010) they measured morphosyntactic attain-
ment by means of a battery of five different tests which were used to create
an overall composite score This battery included a GJT but also an oral nar-
ration task two word order tasks and a gender assignment task Granena and
Long (2013) concluded that the conflicting results of the studies were mostly
due to methodological differences concerning sampling and sample sizes as
well as the aptitude tests procedures and outcome measures employed
Aptitude as measured by tests such as the MLAT or the LLAMA loosely
based on the MLAT may be important in morphosyntactic attainment when
this is assessed by means of tasks that focus on language forms and language
correctness such as GJTs However findings may be different when L2 attain-
ment is measured by means of oral production tasks or other meaning-based
tasks that do not call for the same analytic andor metalinguistic abilities that
also characterize explicit language aptitude measures
The issue of whether a cognitive factor such as language aptitude plays a role
in early L2 acquisition has relevant implications for SLA theory and practice
The rationale behind DeKeyserrsquos (2000) claim in support of a differential role
of aptitude in child and adult L2 acquisition was Bley-Vromanrsquos (1988 1990)
Fundamental Difference Hypothesis according to which there is a qualitative
difference between the learning mechanisms of child and adult L2 learners
younger learners learn mostly implicitly using domain-specific mechanisms
whereas older learners learn mostly explicitly using problem-solving or
domain-general mechanisms and therefore have to rely more on language
aptitude As a result of this qualitative difference between child and adult L2
learning DeKeyser predicted that aptitude would have a differential role in the
two populations If instead language aptitude is found to be equally related to
ultimate attainment in both child and adult L2 learners this could suggest that
child and adult L2 learners are more similar than predicted so far in terms of L2
learning processes
AIMS OF THE PRESENT STUDY
This study focused on the interplay of aptitude and age by investigating the
role of aptitude in early childhood morphosyntactic acquisition in a naturalistic
learning context Unfortunately there is no consensus in the literature regard-
ing a cut-off point between early and late L2 learners and different theoretic-
ally andor empirically motivated cut-offs have been suggested age 12 years
(eg Abrahamsson and Hyltenstam 2008) age 15 years (Johnson and Newport
1989 DeKeyser 2000) and age 18 years (eg DeKeyser et al 2010) Unlike
previous research the present study set out to investigate the role of aptitude
486 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
in early childhood acquisition exclusively defined as L2 learning that started in
the age range between 3 and 6 years Age 3 years was considered a reasonable
choice for dividing simultaneous (2L1) and sequential bilinguals (L2) based on
Meisel (2001 2009) Successive or sequential bilinguals learn one language
first in a given context and the L2 later and in a different context whereas
simultaneous bilinguals learn the two languages since birth and in the same
context According to Meisel (2009) successive language acquisition in early
childhood (as early as age 3 years) can be affected by early maturational
changes which justifies the classification of these sequential bilinguals as
child L2 learners
While there is considerable research on early L2 learners there is not much
research on early childhood learners whose L2 onset began after age 3 years
but before age 6 years From the point of view of learning mechanisms this is a
relevant L2 learner population since they are not expected to be fundamen-
tally different in terms of learning mechanisms from simultaneous bilinguals or
NSs Paradis (2009) for example discusses age 6 years in terms of changes in
explicitndashimplicit learning mechanisms since metalinguistic abilities start de-
veloping around that age Although early childhood L2 learners are not ex-
pected to be fundamentally different in terms of learning mechanisms from
simultaneous bilinguals or NSs they may crucially differ in linguistic attain-
ment and degree of variability (Meisel 2009) In addition early childhood
learners can be expected to be a more homogeneous population with respect
to variables such as type of L2 input available (eg interactional support and
scaffolding patterns) and other social factors that could play a role in
acquisition
The above considerations led to the following research questions
1 Is language aptitude related to early childhood L2 learnersrsquo ultimate mor-
phosyntactic attainment
2 If so is language aptitude equally related to ultimate morphosyntactic
attainment as measured by different language tests andor structures
METHODOLOGY
Participants
A total of 50 Chinese L1ndashSpanish L2 sequential bilinguals (42 per cent males
and 58 per cent females) participated in the study They were all long-term
residents in Madrid (Spain) with no less than a high-school diploma (80 per
cent of them had or were studying a college degree) who arrived in the coun-
try in early childhood (ie between ages 3 and 6 years) or were born in Spain1
In either case these early L2 learners had been born to Chinese-speaking
parents who did not speak the L2 at home as the participants themselves
reported in the biographical questionnaire (see Supplementary Appendix A)
As a result even those early L2 learners who had been born in Spain were not
G GRANENA 487
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
sciences was probably of above-average analytic ability but that his true skills
had not been captured by the aptitude measure used the Words-in-Sentences
subtest in the Modern Language Aptitude Test (MLAT Carroll and Sapon
1959) On the basis of these results DeKeyser concluded that above-average
analytic abilities are required to reach near-native levels in the L2
DeKeyser et al (2010) replicated the findings in DeKeyser (2000) with
Russian speakers of L2 English and L2 Hebrew in the USA and Israel respect-
ively In these two parallel cross-linguistic studies aptitude [operationalized as
L1 verbal aptitude and measured by a test comparable with the verbal
Scholastic Aptitude Test (SAT)] was significantly correlated with ultimate at-
tainment for the adult learners but not for the early learners Specifically the
significant correlation was found for the 18ndash40 years age of acquisition range
but not for the age of acquisition lt18 years group The correlation in the age of
acquisition gt40 years group was not significant either which was interpreted
as the result of factors related to aging
While language aptitude was not related to ultimate morphosyntactic attain-
ment among early acquirers in the studies by DeKeyser (2000) and DeKeyser
et al (2010) Abrahamsson and Hyltenstam (2008) and Bylund et al (2012)
found aptitude effects in pre-pubescent L2 learnersrsquo ultimate attainment who
were first exposed to the L2 before age 12 years In Abrahamsson and
Hyltenstam (2008) there was a significant and moderately strong relationship
between aptitude and ultimate attainment in the early L2-learner group
In this group 72 per cent of the learners who performed within the native
speaker (NS) range had high aptitude operationalized as being above
the grouprsquos average level of aptitude In the group of late learners the four
learners who were able to score within the NS range were all also above
average in terms of aptitude In Bylund et al (2012) aptitude was reported
as the only significant predictor for nativelike performance in both the L1
and the L2
Like DeKeyser (2000) and DeKeyser et al (2010) Abrahamsson and
Hyltenstam (2008) and Bylund et al (2012) relied on grammaticality judge-
ments to measure ultimate attainment Abrahamsson and Hyltenstam (2008)
combined the scores of two different GJT modalities auditory and written
while Bylund et al (2012) used an untimed cloze test and an auditory GJT
that gave participants a time frame of 10 s to respond after the end of each
sentence Unlike DeKeyser (2000) and DeKeyser et al (2010) where aptitude
was operationalized as L1 verbal ability Abrahamsson and Hyltenstam and
Bylund et al used a non-L1-based omnibus test of aptitude the Swansea
Language Aptitude Tests (LAT Meara et al 2003) a combination of five dif-
ferent subtests phonetic memory analytical ability grammatical inferencing
sound recognition and soundndashsymbol correspondence On the basis of the
results of their study Abrahamsson and Hyltenstam (2008) concluded that
lsquoDeKeyserrsquos claim that language aptitude lsquolsquoonly plays a role for adult learnersrsquorsquo
should be modified to state that language aptitude plays not only a crucial role
for adult learners but also a certain role for child learnersrsquo (p 499)
G GRANENA 485
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
However another study Granena and Long (2013) did not find an effect of
aptitude in either early or late learnersrsquo morphosyntactic attainment Like
Abrahamsson and Hyltenstam (2008) they used a non-L1-based omnibus
measure of aptitude the latest version of the LAT the LLAMA aptitude test
(Meara 2005) but unlike Abrahamsson and Hyltenstam (2008) DeKeyser
(2000) and DeKeyser et al (2010) they measured morphosyntactic attain-
ment by means of a battery of five different tests which were used to create
an overall composite score This battery included a GJT but also an oral nar-
ration task two word order tasks and a gender assignment task Granena and
Long (2013) concluded that the conflicting results of the studies were mostly
due to methodological differences concerning sampling and sample sizes as
well as the aptitude tests procedures and outcome measures employed
Aptitude as measured by tests such as the MLAT or the LLAMA loosely
based on the MLAT may be important in morphosyntactic attainment when
this is assessed by means of tasks that focus on language forms and language
correctness such as GJTs However findings may be different when L2 attain-
ment is measured by means of oral production tasks or other meaning-based
tasks that do not call for the same analytic andor metalinguistic abilities that
also characterize explicit language aptitude measures
The issue of whether a cognitive factor such as language aptitude plays a role
in early L2 acquisition has relevant implications for SLA theory and practice
The rationale behind DeKeyserrsquos (2000) claim in support of a differential role
of aptitude in child and adult L2 acquisition was Bley-Vromanrsquos (1988 1990)
Fundamental Difference Hypothesis according to which there is a qualitative
difference between the learning mechanisms of child and adult L2 learners
younger learners learn mostly implicitly using domain-specific mechanisms
whereas older learners learn mostly explicitly using problem-solving or
domain-general mechanisms and therefore have to rely more on language
aptitude As a result of this qualitative difference between child and adult L2
learning DeKeyser predicted that aptitude would have a differential role in the
two populations If instead language aptitude is found to be equally related to
ultimate attainment in both child and adult L2 learners this could suggest that
child and adult L2 learners are more similar than predicted so far in terms of L2
learning processes
AIMS OF THE PRESENT STUDY
This study focused on the interplay of aptitude and age by investigating the
role of aptitude in early childhood morphosyntactic acquisition in a naturalistic
learning context Unfortunately there is no consensus in the literature regard-
ing a cut-off point between early and late L2 learners and different theoretic-
ally andor empirically motivated cut-offs have been suggested age 12 years
(eg Abrahamsson and Hyltenstam 2008) age 15 years (Johnson and Newport
1989 DeKeyser 2000) and age 18 years (eg DeKeyser et al 2010) Unlike
previous research the present study set out to investigate the role of aptitude
486 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
in early childhood acquisition exclusively defined as L2 learning that started in
the age range between 3 and 6 years Age 3 years was considered a reasonable
choice for dividing simultaneous (2L1) and sequential bilinguals (L2) based on
Meisel (2001 2009) Successive or sequential bilinguals learn one language
first in a given context and the L2 later and in a different context whereas
simultaneous bilinguals learn the two languages since birth and in the same
context According to Meisel (2009) successive language acquisition in early
childhood (as early as age 3 years) can be affected by early maturational
changes which justifies the classification of these sequential bilinguals as
child L2 learners
While there is considerable research on early L2 learners there is not much
research on early childhood learners whose L2 onset began after age 3 years
but before age 6 years From the point of view of learning mechanisms this is a
relevant L2 learner population since they are not expected to be fundamen-
tally different in terms of learning mechanisms from simultaneous bilinguals or
NSs Paradis (2009) for example discusses age 6 years in terms of changes in
explicitndashimplicit learning mechanisms since metalinguistic abilities start de-
veloping around that age Although early childhood L2 learners are not ex-
pected to be fundamentally different in terms of learning mechanisms from
simultaneous bilinguals or NSs they may crucially differ in linguistic attain-
ment and degree of variability (Meisel 2009) In addition early childhood
learners can be expected to be a more homogeneous population with respect
to variables such as type of L2 input available (eg interactional support and
scaffolding patterns) and other social factors that could play a role in
acquisition
The above considerations led to the following research questions
1 Is language aptitude related to early childhood L2 learnersrsquo ultimate mor-
phosyntactic attainment
2 If so is language aptitude equally related to ultimate morphosyntactic
attainment as measured by different language tests andor structures
METHODOLOGY
Participants
A total of 50 Chinese L1ndashSpanish L2 sequential bilinguals (42 per cent males
and 58 per cent females) participated in the study They were all long-term
residents in Madrid (Spain) with no less than a high-school diploma (80 per
cent of them had or were studying a college degree) who arrived in the coun-
try in early childhood (ie between ages 3 and 6 years) or were born in Spain1
In either case these early L2 learners had been born to Chinese-speaking
parents who did not speak the L2 at home as the participants themselves
reported in the biographical questionnaire (see Supplementary Appendix A)
As a result even those early L2 learners who had been born in Spain were not
G GRANENA 487
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
However another study Granena and Long (2013) did not find an effect of
aptitude in either early or late learnersrsquo morphosyntactic attainment Like
Abrahamsson and Hyltenstam (2008) they used a non-L1-based omnibus
measure of aptitude the latest version of the LAT the LLAMA aptitude test
(Meara 2005) but unlike Abrahamsson and Hyltenstam (2008) DeKeyser
(2000) and DeKeyser et al (2010) they measured morphosyntactic attain-
ment by means of a battery of five different tests which were used to create
an overall composite score This battery included a GJT but also an oral nar-
ration task two word order tasks and a gender assignment task Granena and
Long (2013) concluded that the conflicting results of the studies were mostly
due to methodological differences concerning sampling and sample sizes as
well as the aptitude tests procedures and outcome measures employed
Aptitude as measured by tests such as the MLAT or the LLAMA loosely
based on the MLAT may be important in morphosyntactic attainment when
this is assessed by means of tasks that focus on language forms and language
correctness such as GJTs However findings may be different when L2 attain-
ment is measured by means of oral production tasks or other meaning-based
tasks that do not call for the same analytic andor metalinguistic abilities that
also characterize explicit language aptitude measures
The issue of whether a cognitive factor such as language aptitude plays a role
in early L2 acquisition has relevant implications for SLA theory and practice
The rationale behind DeKeyserrsquos (2000) claim in support of a differential role
of aptitude in child and adult L2 acquisition was Bley-Vromanrsquos (1988 1990)
Fundamental Difference Hypothesis according to which there is a qualitative
difference between the learning mechanisms of child and adult L2 learners
younger learners learn mostly implicitly using domain-specific mechanisms
whereas older learners learn mostly explicitly using problem-solving or
domain-general mechanisms and therefore have to rely more on language
aptitude As a result of this qualitative difference between child and adult L2
learning DeKeyser predicted that aptitude would have a differential role in the
two populations If instead language aptitude is found to be equally related to
ultimate attainment in both child and adult L2 learners this could suggest that
child and adult L2 learners are more similar than predicted so far in terms of L2
learning processes
AIMS OF THE PRESENT STUDY
This study focused on the interplay of aptitude and age by investigating the
role of aptitude in early childhood morphosyntactic acquisition in a naturalistic
learning context Unfortunately there is no consensus in the literature regard-
ing a cut-off point between early and late L2 learners and different theoretic-
ally andor empirically motivated cut-offs have been suggested age 12 years
(eg Abrahamsson and Hyltenstam 2008) age 15 years (Johnson and Newport
1989 DeKeyser 2000) and age 18 years (eg DeKeyser et al 2010) Unlike
previous research the present study set out to investigate the role of aptitude
486 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
in early childhood acquisition exclusively defined as L2 learning that started in
the age range between 3 and 6 years Age 3 years was considered a reasonable
choice for dividing simultaneous (2L1) and sequential bilinguals (L2) based on
Meisel (2001 2009) Successive or sequential bilinguals learn one language
first in a given context and the L2 later and in a different context whereas
simultaneous bilinguals learn the two languages since birth and in the same
context According to Meisel (2009) successive language acquisition in early
childhood (as early as age 3 years) can be affected by early maturational
changes which justifies the classification of these sequential bilinguals as
child L2 learners
While there is considerable research on early L2 learners there is not much
research on early childhood learners whose L2 onset began after age 3 years
but before age 6 years From the point of view of learning mechanisms this is a
relevant L2 learner population since they are not expected to be fundamen-
tally different in terms of learning mechanisms from simultaneous bilinguals or
NSs Paradis (2009) for example discusses age 6 years in terms of changes in
explicitndashimplicit learning mechanisms since metalinguistic abilities start de-
veloping around that age Although early childhood L2 learners are not ex-
pected to be fundamentally different in terms of learning mechanisms from
simultaneous bilinguals or NSs they may crucially differ in linguistic attain-
ment and degree of variability (Meisel 2009) In addition early childhood
learners can be expected to be a more homogeneous population with respect
to variables such as type of L2 input available (eg interactional support and
scaffolding patterns) and other social factors that could play a role in
acquisition
The above considerations led to the following research questions
1 Is language aptitude related to early childhood L2 learnersrsquo ultimate mor-
phosyntactic attainment
2 If so is language aptitude equally related to ultimate morphosyntactic
attainment as measured by different language tests andor structures
METHODOLOGY
Participants
A total of 50 Chinese L1ndashSpanish L2 sequential bilinguals (42 per cent males
and 58 per cent females) participated in the study They were all long-term
residents in Madrid (Spain) with no less than a high-school diploma (80 per
cent of them had or were studying a college degree) who arrived in the coun-
try in early childhood (ie between ages 3 and 6 years) or were born in Spain1
In either case these early L2 learners had been born to Chinese-speaking
parents who did not speak the L2 at home as the participants themselves
reported in the biographical questionnaire (see Supplementary Appendix A)
As a result even those early L2 learners who had been born in Spain were not
G GRANENA 487
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
in early childhood acquisition exclusively defined as L2 learning that started in
the age range between 3 and 6 years Age 3 years was considered a reasonable
choice for dividing simultaneous (2L1) and sequential bilinguals (L2) based on
Meisel (2001 2009) Successive or sequential bilinguals learn one language
first in a given context and the L2 later and in a different context whereas
simultaneous bilinguals learn the two languages since birth and in the same
context According to Meisel (2009) successive language acquisition in early
childhood (as early as age 3 years) can be affected by early maturational
changes which justifies the classification of these sequential bilinguals as
child L2 learners
While there is considerable research on early L2 learners there is not much
research on early childhood learners whose L2 onset began after age 3 years
but before age 6 years From the point of view of learning mechanisms this is a
relevant L2 learner population since they are not expected to be fundamen-
tally different in terms of learning mechanisms from simultaneous bilinguals or
NSs Paradis (2009) for example discusses age 6 years in terms of changes in
explicitndashimplicit learning mechanisms since metalinguistic abilities start de-
veloping around that age Although early childhood L2 learners are not ex-
pected to be fundamentally different in terms of learning mechanisms from
simultaneous bilinguals or NSs they may crucially differ in linguistic attain-
ment and degree of variability (Meisel 2009) In addition early childhood
learners can be expected to be a more homogeneous population with respect
to variables such as type of L2 input available (eg interactional support and
scaffolding patterns) and other social factors that could play a role in
acquisition
The above considerations led to the following research questions
1 Is language aptitude related to early childhood L2 learnersrsquo ultimate mor-
phosyntactic attainment
2 If so is language aptitude equally related to ultimate morphosyntactic
attainment as measured by different language tests andor structures
METHODOLOGY
Participants
A total of 50 Chinese L1ndashSpanish L2 sequential bilinguals (42 per cent males
and 58 per cent females) participated in the study They were all long-term
residents in Madrid (Spain) with no less than a high-school diploma (80 per
cent of them had or were studying a college degree) who arrived in the coun-
try in early childhood (ie between ages 3 and 6 years) or were born in Spain1
In either case these early L2 learners had been born to Chinese-speaking
parents who did not speak the L2 at home as the participants themselves
reported in the biographical questionnaire (see Supplementary Appendix A)
As a result even those early L2 learners who had been born in Spain were not
G GRANENA 487
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
significantly exposed to the L2 until the beginning of preschool at age 3 years
Until that age they were primarily exposed to Chinese and therefore were
considered sequential not simultaneous bilinguals Specifically 24 of the
early learners started learning the L2 at age 3 years five of them started at
age 4 years 11 at age 5 years and 10 at age 6 years Most of them (88 per cent)
self-reported being Spanish-dominant and the rest (12 per cent) reported being
equally dominant in both Spanish and Chinese These six participants had
started learning the L2 at age 3 (n = 1) 4 (n = 2) 5 (n = 2) and 6 (n = 1) years
Participants were informally screened into the study via a telephone inter-
view conducted by the researcher in order to make sure that participants were
advanced L2 speakers and to exclude those L2 learners who might have been
living in a linguistic ghetto despite having a long length of residence The
inclusion criterion was a score of at least 4 on a 5-point scale that rated their
degree of nativelike pronunciation 5 = Native or near-native pronunciation
no foreign accent 4 = Generally good pronunciation but with occasional non-
native sounds slight foreign accent pronunciation does not interfere with
comprehensibility 3 = Frequent use of non-native sounds noticeable foreign
accent pronunciation occasionally impedes comprehensibility 2 = Generally
poor use of nativelike sounds strong foreign accent pronunciation frequently
impedes comprehensibility 1 = Very strong foreign accent definitely non-
native The average screening score in the group was 48 (SD = 040)
A group of 20 NSs of Spanish (50 per cent males and 50 per cent females)
born in Madrid and with no less than a high school diploma served as controls
Their average age at testing was 2735 years (SD = 518) Table 1 summarizes
the information regarding the L2 learnersrsquo age at testing age of onset of L2
learning length of residence and percentage of Spanish (L2) and Chinese (L1)
use daily The percentages of language use were elicited by means of a question
that asked participants to rank the languages they used on a regular day
indicating an approximate percentage of use
Table 1 L2 learnersrsquo information
Early L2 learners (n = 50)
M Range
Age at testing 2238 (445) 18ndash33
Age of onset 414 (123) 3ndash6
Length of residence 1788 (449) 11ndash28
Percentage of daily L1 use 2850 (1543) 5ndash80
Percentage of daily L2 use 6980 (1552) 40ndash100
Standard deviations appear in parentheses
488 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Target structures
A variety of target structures in Spanish known to be notoriously difficult for
Spanish learners were investigated In total there were six structures Three of
them involved grammatical agreement relations (i) nounndashadjective gender
agreement (ii) subjectndashverb agreement and (iii) nounndashadjective number
agreement These three structures were referred to as [+ agreement] struc-
tures The other three structures made essential contributions to meaning
(iv) subjunctive mood (v) perfectiveimperfective aspect and (vi) passives
with serestar These three structures were referred to as [ agreement]
structures
Instruments and procedures
Participants were administered a speeded-response and a non-speeded-re-
sponse auditory GJT as part of a larger test battery that was presented following
a balanced Latin square design A pool of items was first created and items
were then randomly assigned into the tests (see Supplementary Appendix B
for samples) SuperLab Pro (Cedrus 2003) was used to administer the tests The
choice of GJT measures was based on Ellisrsquo 2005 psychometric study also
replicated by Bowles (2011) according to which tests that do not allow time
to plan responses (ie online measures) may be tapping into implicit language
knowledge whereas tests that allow time to think may be tapping into explicit
language knowledge Participants also completed the LLAMA aptitude test
battery (Meara 2005) and a detailed biographical questionnaire
Speeded-response auditory GJT (k = 60)
The speeded-response GJT was a computer-delivered test with sentences pre-
sented aurally Participants indicated whether each sentence was grammatical
or ungrammatical by pressing a response button within a fixed time-limit
They were asked to press a key as soon as an error was detected in the sen-
tence Once participants pressed a key the computer automatically moved on
to the next sentence without a pause Following Ellis (2005) the time-limit for
each item was established on the basis of NSsrsquo average response time in a pilot
study (n = 10) Following Ellis as well an additional 20 per cent of the time
taken for each sentence was added to allow for the slower processing speed of
L2 learners The time allowed for judging each sentence ranged between
340872 (34 s) to 1004592 ms (10 s) (M = 580798 SD = 100076) from the
onset of the sentence In terms of target structure NSsrsquo longest average re-
sponse times were on aspectual contrasts (M = 536509 SD = 115664) fol-
lowed by gender agreement (M = 510260 SD = 47169) the passive
(M = 498820 SD = 43240) person agreement (M = 489222 SD = 60858)
number agreement (M = 469173 SD = 84426) and the subjunctive
(M = 400005 SD = 71431)
G GRANENA 489
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
which measures the rank-order stability of individualsrsquo scores on different
items of the test was 92
Non-speeded-response auditory GJT (k = 60)
The non-speeded-response GJT was a computer-delivered test with sentences
presented aurally Participants were required to indicate whether each sen-
tence was grammatical or ungrammatical by pressing a response button
Unlike its time-pressured counterpart this test presented each sentence
twice before participants were allowed to provide a response Following
DeKeyser (2000) and DeKeyser et al (2010) each sentence was played
twice with a 3-s interval between the repetitions and a 6-s interval between
sentence pairs
Each item was scored dichotomously as correctincorrect and percentage
accuracy scores were calculated for grammatical and ungrammatical items
overall The internal consistency of the test according to Cronbachrsquos alpha
was 89
Language aptitude test (k = 90)
The language aptitude test was the LLAMA (Meara 2005) the most recent
version of the LAT (Meara et al 2003) The LLAMA is a computer-based ap-
titude test battery that grew out of a series of projects carried out at the
University of Wales Swansea The tests which have an exploratory nature
take approximately 25 min Each subtest is individually and automatically
scored Although largely based on the MLAT the LLAMA tests are described
as being language-independent unlike the MLAT They rely on picture stimuli
and verbal materials adapted from a British-Columbian indigenous language
and a Central-American language The LLAMA includes four sub-tests
LLAMA B a test of vocabulary learning LLAMA D a test of sound recognition
that requires previously heard sound sequences to be identified in new se-
quences LLAMA E a test of soundndashsymbol associations and LLAMA F a
test of grammatical inferencing With the exception of sound recognition
(LLAMA D) the sub-tests include default study phases that last between 2
and 5 min After the study phase test-takers are required to respond to a
series of items with no time pressure The score for each of the LLAMA
sub-tests ranges between 0 and 100 (LLAMA B E and F) and between 0
and 75 (LLAMA D) Feedback is provided after each response in the form of
an acoustic signal
The reliability of the LLAMA test (k = 90) in terms of internal consistency
according to Cronbachrsquos alpha was 77 (an acceptable research standard is
considered to be 70 according to Nunnally and Bernstein 1994)
490 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
RESULTS
The average raw aptitude score in the early childhood learner group was 6235
(SD = 1124) The highest scoring participant obtained 8250 and the lowest
scoring participant obtained 4125 (the maximum score possible was 100)
In the NS control group the average aptitude score was 5763 (SD = 1104)
The highest score was 75 and the lowest 30 The distribution of scores in both
groups was normal according to one-sample K-S tests (p = 945 and p = 895
respectively) An overall aptitude composite score was computed for each
group by converting each of the LLAMA subtest scores into z-scores and
adding them up Although the early learners scored descriptively higher
than the NSs as can be observed in Figure 1 there were no statistically sig-
nificant differences between the two groups according to an independent-
samples t-test [t(68) =1648 p = 104] The equality of variance assumption
(Levenersquos test) was met (p = 757) The size of the effect according to Cohenrsquos d
was 42 (medium) Given the lack of a significant difference the two groups
were considered comparable with regard to aptitude level2
Table 2 displays the descriptive statistics for the grammaticality judgements
on the speeded- and non-speeded-response tests as well as a breakdown of
Speaker Group
Early BilingualsNative Speakers
LL
AM
A te
st m
ean
sco
re
9000
8000
7000
6000
5000
4000
3000
Language Aptitude
Figure 1 Distribution of language aptitude scores by group
G GRANENA 491
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Table
2
Des
crip
tive
stati
stic
s(m
ean
san
dst
an
dard
dev
iati
ons)
for
gram
mati
cali
tyju
dge
men
ts
NS
con
trols
Earl
yL2
learn
ers
[+A
gre
em
en
t][
Agre
em
en
t]Tota
l[+
Agre
em
en
t][
Agre
em
en
t]Tota
l
Speeded
GJT
901
(93
)898
(96
)900
(86
)737
(119
)786
(103
)762
(100
)
Non
-speeded
GJT
937
(67
)922
(89
)930
(63
)759
(142
)829
(82
)794
(97
)
Tota
l919
(67
)91
(77
)747
(121
)807
(82
)
Sta
ndard
devia
tion
sappear
inpare
nth
ese
s
492 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
scores according to type of structure [+ agreement] The three structures
labeled as [+ agreement] were gender agreement number agreement and
subjectndashverb agreement whereas the three structures labelled as [ agree-
ment] were aspect contrasts the subjunctive and the passive (see section on
lsquoTarget Structuresrsquo for more information) As can be seen the early L2 learners
scored lower than the NS controls on the two types of measures and structures
The distribution of scores in each of the groups and for each type of meas-
ure and structure was normal according to one-sample K-S tests (pgt 05) All
between-group differences in overall scores were statistically significant and
associated with a large effect size according to Cohenrsquos d speeded-response
GJT [t(68) = 4175 plt 001 d = 148] non-speeded-response GJT
[t(68) = 4175 plt 001 d = 166] [+ agreement] structures [t(68) = 4175
plt 001 d = 176] and [ agreement] structures [t(68) = 4175 plt 001
d = 129] The equality of variance assumption (Levenersquos test) was met in all
the analyses (pgt 05)
At the within-subjects level the early L2 learners scored significantly lower
on [+ agreement] than [ agreement] structures in both GJTs speeded and
non-speeded [t(49) =3625 plt 001 and t(49) =3944 plt 001] These re-
sults were associated with a medium effect size (d = 044 and d = 060 respect-
ively) In addition L2 learnersrsquo scores on [+ agreement] structures displayed
higher inter-individual variability in both tests as shown by the higher
standard deviations Finally while the correlation between age of onset and
[+ agreement] structures reached significance (r =28 p = 048) the corres-
ponding correlation for [ agreement] structures did not (r =22 p = 123)
suggesting that the strength of the relationship between age of onset and
[+ agreement] structures was slightly stronger than between age of onset
and [ agreement] structures NS controls on the other hand scored descrip-
tively higher on structures involving agreement but unlike in the L2 learner
group in the control group there were no significant differences between the
two types of structures in either test speeded-response GJT [t(19) = 219
p = 829] or non-speeded-response GJT [t(19) = 755 p = 459] The size of the
effect was small in both cases (d = 003 and d = 019)
In order to investigate the role of aptitude in the early L2 learner group
a repeated measures analysis of variance was run with two within-subjects
factors measure (speeded and non-speeded) and structure ([+ agreement] and
[ndash agreement]) and language aptitude as a covariate The results of the multi-
variate tests revealed a significant three-way interaction between type of meas-
ure type of structure and aptitude [F(1 48) = 5286 p = 026 Z2p = 1013
= 899] indicating that aptitude moderated the scores on the two types of
structures differently depending on the type of test (speeded or non-speeded)
Follow-up analyses in the form of pairwise comparisons confirmed that apti-
tude played a role in the non-speeded-response GJT as indicated by a signifi-
cant interaction between type of structure and aptitude [F(1 48) = 4194
p = 046 Z2p = 082 = 918] This interaction was not significant in the case
of the speeded-response GJT [F(1 48) = 006 p = 937 Z2p = 000 = 1000]
G GRANENA 493
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
A summary of the correlations between aptitude and type of structure in each
of the tests is displayed in Table 3
In the NS control group the repeated measures general linear model did not
yield any significant results in terms of language aptitude The interactions
between aptitude and type of measure or aptitude and type of structure
were not significant [F(1 18) = 002 p = 961 Z2p = 000 = 1000 and
F(1 18) = 531 p = 476 Z2p = 029 = 971] The three-way interaction be-
tween aptitude type of measure and type of structure was not significant
either [F(1 18) = 1178 p = 292 Z2p = 061 = 939] Table 4 displays the
simple correlations between aptitude and each of the dependent variables in
the NS control group As can be seen none of the correlations was significant
and no clear patterns could be observed regarding the r values either even
though the magnitude of the effect for the speeded-response GJT with struc-
tures that did not involve agreement [ agreement] was similar to that
observed in the L2-learner group
A last set of analyses was conducted to examine whether language aptitude
was a necessary condition for early childhood L2 learners to score within NS
range as determined by the lowest-scoring NS in each of the measures
Participants were first divided into high- and low-aptitude according to the
median-split method According to this method 24 L2 learners and 10 NSs had
high aptitude (M = 7188 SD = 547 and M = 6625 SD = 421 respectively)
whereas 26 L2 learners and 10 NSs had low aptitude (M = 5356 SD = 726
and M = 490 SD = 862 respectively)
GJT scores were then regressed on age of onset as shown in Figures 2 and 3
The aptitude dimension was added in order to distinguish between high- and
low-aptitude participants As can be seen there were low-aptitude L2 learners
who scored within the NS range in the two tests Overall however the per-
centage of high-aptitude L2 learners with GJT scores within the NS range was
higher than that of low-aptitude L2 learners in both GJTs the speeded (583
vs 385 per cent)4 and the non-speeded (75 vs 46 per cent) In addition the
only five L2 learners who were able to score within the NS range in the two
tests examined were high-aptitude individuals
To summarize language aptitude as measured by an omnibus test of apti-
tude (the LLAMA test) was significantly related to variation in early childhood
Table 3 Correlations between aptitude and scores in early L2 learner group(n = 50)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
Speeded GJT[ agreement]
Speeded GJT[+ agreement]
Language aptitude 26 40 24 20
067 005 096 171
plt 01
494 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
L2 learnersrsquo attainment Specifically aptitude was related to early learnersrsquo
scores on structures involving grammatical agreement relations in a non-
speeded-response GJT Although aptitude played a role in the early L2 learner
group it was not a necessary condition for early learners to attain a score
within the NS range in each of the tests when the tests were considered
separately It was necessary however for nativelike performance across-
the-board in both of the tests examined Finally no relationship was found
between aptitude and language attainment in the NS control group
DISCUSSION
This study set out to investigate the interplay between aptitude and age in
early morphosyntactic acquisition specifically in a group of early childhood L2
learners with ages of onset between 3 and 6 years Although these learners had
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Speeded Auditory GJT
High-Aptude Low-Aptude
Figure 2 Speeded auditory GJT scores as a function of age of onset with theaptitude dimension added
Table 4 Correlations between aptitude and scores in NS control group(n = 20)
Non-speededGJT[ agreement]
Non-speededGJT[+ agreement]
SpeededGJT[ agreement]
SpeededGJT[+ agreement]
Language aptitude 10 21 25 01
677 378 293 972
G GRANENA 495
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
started learning the L2 at a very early age their language attainment was
significantly lower than NSsrsquo and characterized by greater inter-individual
variability These results indicate that the acquisition of morphosyntax for
certain L1ndashL2 pairings (Chinese L1ndashSpanish L2 in this case) may be affected
even when the L2 is acquired as early as age 3 years Chinese and Spanish are
two typologically distant languages with very different inflectional paradigms
(uniform vs complex) Unlike Spanish Chinese is an isolating language in
which almost every word consists of a single morpheme There is no nounndash
adjective gender or number agreement and no subjectndashverb agreement or
mood (indicativesubjunctive) Chinese has a marker for the passive but no
alternation between two copula verbs like Spanish Finally Chinese encodes
aspect in the form of free-standing morphemes but there is no one-to-one
mapping between aspect contrasts in the two languages These findings are
similar to some of the findings reported by Meisel (2009) who claimed that in
certain areas of morphosyntax child L2 acquisition can resemble adult L2
acquisition and differ from L1 acquisition In order to explain these empirical
findings he proposed a modified version of the Critical Period Hypothesis as
suggested by Lenneberg (1967) and others His modifications (eg Meisel
2008) concerned the age at which maturation starts affecting L2 morphosyntax
(around age 4 years) much earlier than the critical age range hypothesized by
Lenneberg (1967) (ie puberty) and the scope of the Critical Period which as
already pointed out by Long (1990) does not equally affect all language do-
mains and grammatical properties
In the current study the target structures where language attainment
was the poorest in the early L2 learner group were the three structures invol-
ving grammatical agreement (gender agreement number agreement and
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7
Mea
n
Acc
urac
y Sc
ore
Age of Onset
Non-speeded Auditory GJT
High-Aptude Low-Aptude
Figure 3 Non-speeded auditory GJT scores as a function of age of onset withthe aptitude dimension added
496 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
subjectndashverb agreement) Early L2 learnersrsquo performance on these structures
was significantly lower than on structures that did not involve grammatical
agreement These results were consistent across the two language measures
investigated (speeded and non-speeded GJT) and differed from the pattern
observed in the NS control group where scores were descriptively higher for
agreement structures and where there were no significant differences between
the two types of target structures These findings provide evidence of the large
range of variation in early childhood acquisition not only across individuals
but also within learners as a function of grammatical structure They also
suggest that certain grammatical properties are affected by early maturational
changes to a larger extent than others Specifically structures in the area of
inflectional morphology (gender number and subjectndashverb agreement) seem
to be particularly sensitive to maturational changes as hypothesized by Meisel
(2009) These are structures that L1 Spanish children acquire very early (ie by
age 3 years) whereas structures such as the subjunctive the passive and
aspect contrasts are acquired later (ie at least age 7 years or later) (see
Lopez Ornat 1994 Montrul 2004)
Meisel (2009) building on Smith and Tsimpli (1995) explains early matur-
ational changes in inflectional morphology as the result of inaccessibility to
parameterized universal grammar (UG) principles specifically inaccessibility to
uninterpretable features of functional categories An alternative non-UG-based
explanation could be that there are very early changes in the capacity for
implicit learning and that this capacity deteriorates with increasing age
(Hoyer and Lincourt 1998) even though it is not lost This would have a
differential impact on the acquisition of grammatical structures Less salient
non-meaning-bearing structures (especially those involving co-occurrence
patterns such as grammatical agreement) could be more affected by changes
in the efficiency of implicit learning mechanisms These are structures to which
infants are extremely sensitive and finely tuned in L1 acquisition but which
become persistent learning problems in L2 acquisition especially when gram-
matical features differ between L1 and L2 as it was the case for the population
investigated since Chinese lacks inflectional morphology
If maturational changes start affecting the acquisition of morphosyntax as
early as age 3 or 4 years one would expect individual differences such as
language aptitude to start playing a compensatory role as early as age 3 or 4
years particularly in those structures that are more affected by maturational
changes The results of this study showed that aptitude is indeed one of the
factors that can account for the high inter-individual variability that was
observed in the early L2 learner data and which contrasted with the high
inter-individual homogeneity observed in the NS control group where apti-
tude did not play any role These findings are in line with Abrahamsson and
Hyltenstam (2008) and give support to their claim that aptitude plays lsquonot only
a crucial role for adult learners but also a certain role for child learnersrsquo (p
499) However they conflict with DeKeyser (2000) DeKeyser et al (2010)
and Granena and Long (2013) where aptitude did not play a role in early L2
G GRANENA 497
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
learnersrsquo morphosyntactic attainment The conflicting results of these studies
as also argued in Granena and Long (2013) may be due to methodological
differences concerning sampling and sample sizes as well as the aptitude tests
procedures and outcome measures employed
In DeKeyser (2000) and DeKeyser et al (2010) the lack of a relationship
between aptitude and ultimate attainment in early learners could have been
due to the restricted range (ie little variation) of scores on both the GJT (eg
all the early learners scored above 90 per cent in DeKeyser 2000) and prob-
ably the aptitude test where scores could have been lower and less varied
than in other aptitude measures This test was administered in the participantsrsquo
L1 which could have led to confounds between participantsrsquo proficiency level
and their cognitive capacity since degree of L1 attrition tends to be inversely
related to age of L2 acquisition (Montrul 2008) For example in DeKeyser
(2000) the highest scorer on the aptitude test (an L1 version of the Words-
in-Sentences MLAT subtest) was the latest arrival (age of arrival = 38 years)
The next highest aptitude scorers were also late arrivals Conversely early
arrivals were not able to score as high as late arrivals perhaps because early
L2 learners do not typically receive formal education in their L1 usually used
in the home environment for conversations with friends and family As a
result they tend to develop poorer L1 literacy skills (reading and writing)
probably affecting language aspects such as vocabulary richness indices
which could play a role in L1 verbal aptitude measures such as the one used
in DeKeyser et al (2010) a test comparable with the verbal SAT a standar-
dized test for most college admissions in the USA
In Granena and Long (2013) on the other hand the lack of a relationship
between aptitude and ultimate attainment seems to have been due to the fact
that morphosyntactic attainment was assessed by means of a combination of
different morphosyntactic measures and not just by means of a GJT This is a
relevant factor when aptitude is measured via tests such as the LLAMA loosely
based on the MLAT because language tests that encourage a focus on language
form and language correction (GJTs) and aptitude tests that are weighted in
favour of explicit processes (eg language analytic ability) may be actually
measuring the same underlying abilities (Long 2007)
This is in fact an explanation that could account for the findings in
Abrahamsson and Hyltenstam (2008) as well as in the present study since
both studies used GJTs as a measure of ultimate attainment and tests loosely
based on the MLAT as a measure of aptitude Abrahamsson and Hyltenstam
(2008) combined the scores of two different GJT modalities an auditory
(online) and a written (offline) with no time pressure and they found a sig-
nificant relationship for aptitude in the early group (r = 70 plt 001) In the
present study that relationship was only significant for the non-speeded-re-
sponse auditory GJT (r = 40 p = 005) a test with offline features since it
allowed participants to listen to each stimulus sentence twice with a 3-s inter-
val between the repetitions and a 6-s interval between sentence pairs
Untimed L2 measures that focus on language correctness such as this one
498 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
may induce learners to approach the task more analytically by placing a great
deal of conscious (ie controlled) attention on sentence structure As a result
L2 learners with higher analytic abilities as measured by tests such as the
LLAMA could be more successful at detecting grammatical errors under un-
timed test conditions especially on those structures with the greatest inter-
individual variability The lack of any significant relationships between lan-
guage attainment and aptitude in the NS control group on the other hand
could be due to the lack of inter-individual variability in NSsrsquo performance on
the language measures employed in combination with the smaller sample size
of the group This means that a relationship between NSsrsquo aptitude and their
language performance might be possible on language measures that allow for
greater variability or that have no ceiling (eg reaction time measures)
Language aptitude in the present study correlated with L2 attainment in an
untimed but not in a timed test This could be interpreted as suggesting that
aptitude is not related to linguistic competence understood as implicit lan-
guage knowledge that can be used automatically However another possibility
is that a different type of aptitude correlates with more spontaneous use of L2
knowledge For example Granena (2013c) found that a type of aptitude
hypothesized to be relevant for implicit learning and processing (Granena
2013a) was related to performance on a word monitoring task an online
meaning-focused task that requires automatic use of language knowledge
The question would be then what type of language and aptitude measures
can provide the most robust evidence for the role of aptitude in L2 acquisition
A tentative answer could be that robust evidence should come from (i) lan-
guage measures considered to be the best means of elicitation of automatic use
of L2 knowledge (eg spontaneous production tasks) and (ii) aptitude meas-
ures tapping into cognitive abilities that are very different from those that the
language measures are likely to engage participants in In other words the
language tests and aptitude tests used should not be measuring the same
abilities This would help prevent confounds arising from the measures them-
selves (ie test effects) and would increase the validity of the findings
Although language aptitude was found to play a role in early childhood
learnersrsquo ultimate attainment the results of the study showed that it was
not a necessary condition to score within the range of NSs a pattern that
was also observed in Abrahamsson and Hyltenstam (2008) The extent to
which aptitude is required to reach nativelikeness could have been a potential
difference between the role of aptitude in early and late acquisition in line
with DeKeyserrsquos (2000) prediction However aptitude does not seem to be
necessary for either early or late L2 learners to perform in a nativelike fashion
at least when a single language measure is considered For example in
Granena (2013b) the late learner with the highest score on an auditory GJT
(a score within the NS range) was in fact a low-aptitude individual These
results suggest that aptitude can moderate L2 attainment at a group level
but that it is not a necessary condition at an individual level in order to
reach nativelike levels on a given language measure Aptitude may be
G GRANENA 499
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
necessary however for nativelike performance across multiple measures even
among very early L2 learners As other studies have shown (eg Abrahamsson
and Hyltenstam 2009 Granena and Long 2013) the use of complex multiple-
task designs and stringent elicitation methods and techniques of analysis is
crucial to understand the scope of age effects and the role of aptitude In the
present study the only five L2 learners who were able to score within the NS
range in the two measures examined were all high-aptitude individuals
Increasing the number of language measures from two to four and operationa-
lizing aptitude as a combination of implicit and explicit cognitive abilities
(Granena 2013a) yielded the same results with no low-aptitude early L2 lear-
ner able to perform like a NS across the whole range of measures Further
research is clearly needed in this area in order to confirm that aptitude plays
more than a lsquocertainrsquo role for child learners thus modifying both DeKeyserrsquos
(2000) claim that aptitude lsquoonly plays [a] role for adult learnersrsquo (p 515) and
Abrahamsson and Hyltenstamrsquos (2008) claim that it lsquoplays not only a crucial
role for adult learners but also a certain role for child learnersrsquo (p 499)
CONCLUSIONS
The aim of the present study was to examine whether language aptitude
played a role in early childhood L2 acquisition and whether this role was
the same across different language measures and structures The findings
demonstrated the presence of an interaction between aptitude type of lan-
guage measure and target structure Aptitude was significantly related to early
learnersrsquo attainment on a non-speeded-response auditory GJT specifically to
target structures involving grammatical agreement These were the structures
for which age effects were the strongest and where variability was the greatest
These results give support to Abrahamsson and Hyltenstamrsquos (2008) claim that
language aptitude plays a role not only in adult SLA but also in child SLA The
fact that such a role was already present in early childhood L2 learners and
that these learners performed significantly lower than NSs as a group can be
interpreted as indicating the existence of very early maturational changes
taking place at least in certain areas of morphosyntax as claimed by Meisel
(2009) However aptitude was only significantly correlated with L2 attain-
ment as measured by a non-speeded-response GJT This was interpreted as
being the result of the type of aptitude investigated weighted in favour of
explicit cognitive processes and the untimed nature of the GJT which induced
learners to approach the task analytically
This study was only able to give a partial picture of early childhood acqui-
sition Other individual differences that are not cognitive and that were not
investigated may also be relevant to account for variability in early learnersrsquo
long-term L2 achievement As in all ex-post-facto designs where pre-existing
groups of individuals are compared there may also be other overlapping vari-
ables that could have influenced the results but that could not be controlled
For example the quality of the input that participants were exposed to in the
500 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
initial stages of the acquisition process was likely to be more homogenous than
in the case of late childhood or late L2 learners However there could have
been differences in terms of use andor type of interlocutor that could have
also explained variability in learning outcomes Overall this study suggests a
more complex picture for L2 acquisition in early childhood than generally
considered in the literature Therefore further research should investigate
other individual differences that may play a role in very early L2 learners as
well as early learners with ages of onset between 7 and 11 in order to compare
the effects of aptitude in early childhood and late childhood acquisition
SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online
ACKNOWLEDGEMENT
This material is based upon work supported by the National Science Foundation under Grant No
1124126
NOTES
1 Nowadays Madrid Barcelona and the
Canary Islands are home to the largest
Chinese communities of Spain Over 80
per cent of the Chinese in Spain come
from the Zhejiang province with smal-
ler numbers from Guangdong and
Fujian
2 A previous study by Granena and Long
(2013) did find significant differences in
language aptitude between early child-
hood L2 learners with ages of onset be-
tween 3 and 6 years late childhood L2
learners adult L2 learners and NSs
Early childhood L2 learners had signifi-
cantly higher aptitude than all the
other groups
3 For partial eta squared (Z2p) a small
effect size is 01 Z2p lt 06 medium is
06 Z2p lt 14 and large is Z2
p 14
4 These percentages correspond to the
early L2 learners who scored within
the NS range in the speeded-response
GJT after removing the NS who scored
3 standard deviations below the mean
of the group The score of this NS (608)
can be seen on the graph
REFERENCES
Abrahamsson N and K Hyltenstam 2008
lsquoThe robustness of aptitude effects in near-
native second language acquisitionrsquo Studies in
Second Language Acquisition 30 481ndash509
Abrahamsson N and K Hyltenstam 2009
lsquoAge of onset and nativelikeness in a se-
cond language Listener perception versus lin-
guistic scrutinyrsquo Language Learning 59
249ndash306
Bley-Vroman R 1988 lsquoThe fundamental char-
acter of foreign language learningrsquo
in W Rutherford and M Sharwood Smith
(eds) Grammar and Second Language Teaching
A Book of Readings Newbury House
pp 133ndash59
Bley-Vroman R 1990 lsquoThe logical problem of
foreign language learningrsquo Linguistic Analysis
20 3ndash49
G GRANENA 501
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Bowles M 2011 lsquoMeasuring implicit and expli-
cit linguistic knowledgersquo Studies in Second
Language Acquisition 33 247ndash71
Bylund E N Abrahamsson and
K Hyltenstam 2012 lsquoDoes first language
maintenance hamper nativelikeness in a
second language A study of ultimate attain-
ment in early bilingualsrsquo Studies in Second
Language Acquisition 34 215ndash41
Carroll J B 1981 lsquoTwenty-five years of re-
search in foreign language aptitudersquo
in K Diller (ed) Individual Differences and
Universals in Language Learning Aptitude
Newbury House pp 83ndash118
Carroll J B and S Sapon 1959 Modern
Language Aptitude Test Form A Psychological
Corporation
Cedrus (2003) SuperLab 307 Cedrus
Corporation
DeKeyser R M 2000 lsquoThe robustness of crit-
ical period effects in second language acquisi-
tionrsquo Studies in Second Language Acquisition 22
499ndash533
DeKeyser R M I Alfi-Shabtay and
D Ravid 2010 lsquoCross-linguistic evidence for
the nature of age-effects in second language
acquisitionrsquo Applied Psycholinguistics 31
413ndash38
Doughty C M Bunting S Campbell
A Bowles and H Haarmann 2007
Development of the High-level Language Aptitude
Battery Technical Report Center for
Advanced Study of Language University of
Maryland College Park
Doughty C S Campbell M Mislevy
M Bunting A Bowles and J Koeth 2010
lsquoPredicting near-native ability The factor struc-
ture and reliability of Hi-LABrsquo in M Prior
Y Watanabe and S Lee (eds) Selected
Proceedings of the 2008 Second Language Research
Forum Cascadilla Proceedings Project pp 10ndash31
Ellis R 2005 lsquoMeasuring implicit and explicit
knowledge of a second language A psycho-
metric studyrsquo Studies in Second Language
Acquisition 27 141ndash72
Granena G 2013a lsquoCognitive aptitudes for
second language learning and the LLAMA
Language Aptitude Testrsquo in G Granena and
M H Long (eds) Sensitive Periods Language
Aptitude and Ultimate L2 Attainment John
Benjamins pp 105ndash29
Granena G 2013b lsquoReexamining the robust-
ness of aptitude in second language acquisi-
tionrsquo in G Granena and M H Long (eds)
Sensitive Periods Language Aptitude and
Ultimate L2 Attainment John Benjamins
pp 179ndash204
Granena G 2013c lsquoIndividual differences in
sequence learning ability and SLA in early
childhood and adulthoodrsquo Language Learning
63 665ndash703
Granena G and M H Long 2013 lsquoAge of
onset length of residence aptitude and ulti-
mate L2 attainment in three linguistic
domainsrsquo Second Language Research 29 311ndash43
Harley B and D Hart 1997 lsquoLanguage apti-
tude and second language proficiency in class-
room learners of different starting agesrsquo Studies
in Second Language Acquisition 19 379ndash400
Hoyer W J and A E Lincourt 1998 lsquoAgeing
and the development of learningrsquo in M
A Stadler and P A Frensch (eds) Handbook
of Implicit Learning Sage pp 445ndash70
Johnson J S and E L Newport 1989
lsquoCritical period effects in second language
learning The influence of maturational state
on the acquisition of English as a second lan-
guagersquo Cognitive Psychology 21 60ndash99
Lenneberg E 1967 Biological Foundations of
Language Wiley
Long M H 1990 lsquoMaturational constraints on
language developmentrsquo Studies in Second
Language Acquisition 12 251ndash85
Long M H 2007 Problems in SLA Erlbaum
Lopez Ornat S 1994 La adquisicion de la lengua
espanola Siglo XXI
Meara P 2005 LLAMA Language Aptitude Tests
Lognostics
Meara P J Milton and N Lorenzo-Dus
2003 Swansea Language Aptitude Tests (LAT)
v20 Lognostics
Meisel J 2001 lsquoThe simultaneous acquisition of
two first languages Early differentiation and
subsequent development of grammarsrsquo
in J Cenoz and F Genesee (eds) Trends in
Bilingual Acquisition John Benjamins pp 11ndash41
Meisel J 2008 lsquoChild second language acquisi-
tion or successive first language acquisitionrsquo
in B Haznedar and E Gavruseva (eds)
Current Trends in Child Second Language
Acquisition A Generative Perspective John
Benjamins pp 55ndash80
502 LANGUAGE APTITUDE AND EARLY L2 ACQUISITION
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from
Meisel J M 2009 lsquoSecond language acquisition
in early childhoodrsquo Zeitschrift fur
Sprachwissenschaft 28 5ndash34
Montrul S 2004 The Acquisition of Spanish
Morphosyntactic Development in Monolingual and
Bilingual L1 Acquisition and Adult L2
Acquisition John Benjamins
Montrul S 2008 Incomplete Acquisition in
Bilingualism Re-examining the Age Factor John
Benjamins
Nunnally J C and I H Bernstein 1994
Psychometric Theory 3rd edn McGraw-Hill Inc
Paradis M 2009 Declarative and Procedural
Determinants of Second Languages John
Benjamins
Skehan P 1989 Individual Differences in Second
Language Learning Arnold
Smith N and I M Tsimpli 1995 The Mind of
a Savant Language Learning and Modularity
Blackwell
G GRANENA 503
at Nipissing U
niversity on October 10 2014
httpapplijoxfordjournalsorgD
ownloaded from