Mayberry Lock 2003 PDF

Brain and Language 87 (2003) 369–384

www.elsevier.com/locate/b&l

Age constraints on first versus second language acquisition:

Evidence for linguistic plasticity and epigenesisq
Rachel I. Mayberry* and Elizabeth Lock
School of Communication Sciences and Disorders, McGill University, 1266 Pine Avenue West, Montreal, Que., Canada H3G 1A8
Accepted 17 April 2003
Abstract
Does age constrain the outcome of all language acquisition equally regardless of whether the language is a first or second one? To
test this hypothesis, the English grammatical abilities of deaf and hearing adults who either did or did not have linguistic experience
(spoken or signed) during early childhood were investigated with two tasks, timed grammatical judgement and untimed sentence to
picture matching. Findings showed that adults who acquired a language in early life performed at near-native levels on a second
language regardless of whether they were hearing or deaf or whether the early language was spoken or signed. By contrast, deaf
adults who experienced little or no accessible language in early life performed poorly. These results indicate that the onset of
language acquisition in early human development dramatically alters the capacity to learn language throughout life, independent of
the sensory-motor form of the early experience.
Ó 2003 Elsevier Science (USA). All rights reserved.
Keywords: Critical period; Grammatical processing; Comprehension; First-language acquisition; Second-language acquisition; Signed language;
ASL; Plasticity; Syntax
1. Introduction (1959) first proposed that language acquisition was re-

lated to brain plasticity. Lenneberg (1967) later mar-
An important question about the nature of language shaled a variety of evidence linking the trajectory of
acquisition is the extent to which age constrains its language acquisition to brain growth curves in early
outcome, otherwise known as a sensitive or critical pe- development. Despite this long history, however, the
riod (CP) for language. The idea that languages must be nature of the postulated critical period for language is
learned in childhood to be learned successfully has been not well understood. Indeed, the existence of a critical
widely held by educators for over a century (Colombo, period for language acquisition remains controversial.
1982). The specific neurolinguistic hypothesis that the In the present study we investigate this important
outcome of language acquisition is tied to brain devel- question with a new approach.
opment has a more recent history. Penfield and Roberts Investigating a possible CP for language requires
identifying situations where the developmental onset of
language acquisition varies naturally. Possible effects on
q
The research reported here was supported by a grant from the the outcome of language acquisition associated with
Natural Sciences and Engineering Research Council of Canada learning languages at various ages can then be mea-
(#171239). Part of the data was collected for an M.Sc. thesis by E.
Lock and funded by a research bursary from the Fond de la recherche
sured. The most common test of the CP hypothesis has
en sante du Quebec through the McGill University Faculty of been spoken, second language (L2) learning because age
Medicine. We thank the participants for volunteering their time, of L2 learning varies widely in the hearing population
Pamela Witcher, Hena Kazmi, and Charlene Chamberlain for help (Birdsong, 1999). A less common situation is the signed
testing participants and Daphne Ducharme, Fred Genesee, Jill language acquisition of individuals who are born deaf
Morford, and Virginia Swisher for helpful comments on earlier
versions of the paper.
(Mayberry, 1994, 2002). We compare the outcome of
*
Corresponding author. Fax: +514-398-8123. these two situations in the present study to probe the
E-mail address: [email protected] (R.I. Mayberry). nature of the postulated CP for language. Specifically,
0093-934X/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0093-934X(03)00137-8
370 R.I. Mayberry, E. Lock / Brain and Language 87 (2003) 369–384
we ask whether the onset of language acquisition in task but not on a grammatical judgement task. Together
early life is related to the subsequent ability to learn any these results indicate that age of acquisition is an im-
other language for the remainder of life, independent of portant factor in the outcome of signed language ac-
the sensory and motor modalities of the first or second quisition. Clearly age constraints on language
languages. Positive evidence of this kind would suggest acquisition are not limited to spoken languages to which
that the postulated CP for language is similar to other we now turn.
biological phenomena whereby early experience orga- The most common method of investigating age con-
nizes the development of a genetically specified system straints on the outcome of language acquisition has been
and its neural underpinnings in an epigenetic fashion to measure the grammatical ability of individuals who
(Changeux, 1985) as we explain below. Before describ- learned a second spoken language at varying ages. Some
ing the present study, we turn to previous research on studies have investigated age constraints on the outcome
age of acquisition effects on the grammatical outcome of of L2 phonological learning but, because the focus of
language acquisition beginning with the case of signed the present study is grammatical ability, we do not dis-
language, followed by spoken language. cuss them here (see Flege, 1999). Several studies re-
Several studies have investigated age of acquisition ported a negative correlation between age of spoken L2
effects on the outcome of American sign language acquisition and L2 grammatical outcome and/or signif-
(ASL). ASL is the most commonly used signed language icant differences in grammatical performance between
in North America but only one of the worldÕs many native and non-native learners. These effects were found
signed languages (Klima & Bellugi, 1979). Signed lan- using a variety of language measures including: sentence
guages are natural languages that have evolved through shadowing (Oyama, 1978), assessment of written tran-
generations of childrenÕs acquisition and adult use by scripts of spoken interviews (Patkowski, 1980), and as-
Deaf communities worldwide (Baynton, 1996; Senghas sessment of tape-recorded interviews (White & Genesee,
& Coppola, 2001). Because they are natural languages 1996). Other studies reported effects for age of acquisi-
independent of spoken languages, signed languages are tion on L2 grammatical outcome using judgement of
neither universal nor gesture codes for speech (Morford grammatical and ungrammatical sentences presented in
& Kegl, 2000). The linguistic architecture of signed either auditory or written forms (Birdsong, 1992; Bird-
language is similar to that of spoken language in that it song & Molis, 2001; Flege, Yeni-Komshian, & Liu,
is characterized by rule-bound form at the levels of 1999; Johnson & Newport, 1991; Johnson, 1992; White
phonology, morphology and syntax, and semantics (for & Genesee, 1996). In most studies the L2 tested was
a review see Emmorey, 2002). Infants exposed to ASL English; French was the L2 in one study (Birdsong,
by their parents acquire it in a fashion and on a time- 1992). The first languages (L1) were Chinese, French,
table akin to hearing childrenÕs acquisition of spoken Italian, Korean, Spanish, or unspecified.1
languages (Chamberlain, Morford, & Mayberry, 2000; Although a negative correlation between age of L2
Lillo-Martin, 1999; Petitto & Marentette, 1991). acquisition and grammatical outcome has been repli-
In the first study of age constraints on ultimate at- cated several times using a variety of language measures
tainment in ASL, Mayberry and Fischer (1989) found across a variety of first languages, controversy remains
significant differences in the narrative shadowing per- as to whether these findings provide positive evidence
formance and lexical error patterns of native signers for the postulated CP for language. Most studies found
(who learned ASL from their deaf parents) compared to L2 grammatical outcome to show a linear function in
that of non-native signers (who learned ASL between relation to age of acquisition; as age of acquisition in-
the ages of 9 and 16). In a second experiment, perfor- creases, L2 grammatical outcome decreases after the age
mance accuracy on ASL sentence shadowing and recall of 8 (e.g., Birdsong & Molis, 2001; Flege et al., 1999;
tasks showed a linear relation to age of acquisition Johnson & Newport, 1989; Oyama, 1978; Patkowski,
(between the ages of birth to 15 years), when length of 1980). However, some researchers have argued that the
ASL experience was a confounding factor. In a third slope of the function between age of acquisition and
experiment controlling length of experience, age of ac- grammatical outcome should be non-linear in nature
quisition continued to show a significant linear relation and stop abruptly at some age coincident with the end of
to performance accuracy and morphological error pat- the CP (e.g., Bialystok & Hakuta, 1999).
terns on a task of complex ASL sentence recall (May- A non-linear function between age of L2 acquisition
berry & Eichen, 1991). Newport (1990) also found age and grammatical outcome was found by Johnson and
of acquisition (from birth to older than 13 years) to Newport (1989), who tested native speakers of Chinese
correlate with ASL ultimate attainment using a com- and Korean with an untimed, grammatical judgement
posite score derived from a battery of expressive and
receptive ASL tests. Finally, Emmorey, Bellugi, Fried- 1
Several studies claiming age of acquisition effects for L2 outcome
erici, and Horn (1995) found native ASL learners to are not cited here because they either did not control for practice effects
outperform non-native learners on a sign monitoring or perform the necessary statistical tests.
R.I. Mayberry, E. Lock / Brain and Language 87 (2003) 369–384 371
task presented auditorally (with a written response) us- (Mayberry, 1994, 2002; Morford & Mayberry, 2000).
ing a variety of English grammatical structures. Age of Babies who hear normally are exposed to spoken lan-
L2 acquisition correlated with grammatical judgement guage from birth nearly without exception. By contrast,
scores between the ages of 3 and 15 (r ¼ :87) but not infants who are born severely and profoundly deaf are
between the ages of 17 and 39 (r ¼ :16). These findings isolated from the language spoken around them by
were interpreted to mean that the ability to acquire any virtue of their deafness. Except in a minority of cases
language, first or second, disappears with increasing where the parents use ASL (i.e., <10%, Schein & Delk,
maturation. Two subsequent studies failed to replicate 1974) the majority of deaf children are not exposed to a
this non-linear function between age of L2 acquisition signed language until older ages, often after enrolling in
and grammatical outcome, however. school. Deaf childrenÕs exposure to accessible language
First, Flege et al. (1999) tested Korean learners of can be delayed for two reasons. Initial detection of
English with procedures similar to those used by John- deafness and the provision of rehabilitation services
son and Newport (1989). Pre-puberty ages of acquisition often occur at ages beyond early childhood. Signed
correlated with grammatical outcome (3–12 years, language may also be withheld from deaf children in the
r ¼ :52) but, importantly, post-puberty ages also cor- belief that doing so will promote speech development.
related with grammatical outcome (13–21 years, For many deaf children, language exposure restricted to
r ¼ :27). In another study, Birdsong and Molis (2001) speech provides insufficiently detailed linguistic input for
used the same methods and English stimuli as Johnson language development to occur spontaneously (Leder-
and Newport (1989) but with native Spanish speakers. berg & Everhart, 1998). This unique circumstance of
Age of L2 acquisition did not correlate with grammat- language acquisition thus permits a novel test of the CP
ical outcome between the ages of 3 and 15. This was for language. Specifically, it allows us to investigate the
primarily due to ceiling effects; many L2 learners per- effects of the onset of first as compared to second lan-
formed within the range of the native English speakers. guage acquisition on grammatical outcome.
Moreover, age of L2 acquisition correlated with gram- We previously investigated the question of whether
matical outcome at ages well beyond childhood, spe- age constraints are greater for the outcome of L1 as
cifically, between the ages of 17 and 44 ðr ¼ :69Þ, compared to L2 (Mayberry, 1993). All participants were
corroborating the findings of Flege et al. (1999). These deaf adults. One group was born with normal hearing
findings are counter to the maturation hypothesis. which they suddenly lost between the ages of 9 and 13
The results of studies investigating the relation be- years due to various viral infections; after becoming
tween age of acquisition and L2 grammatical outcome deaf, they learned ASL as an L2. The contrast group
are thus inconclusive with respect to the existence of a was born deaf and acquired ASL at matched ages (9–13
CP for language acquisition. Declines in L2 grammatical years) but after little or no prior accessible language
performance associated with increasing age of L2 ac- exposure; their ASL acquisition was clearly a case of
quisition after age 8 have been reported in several significantly postponed first-language acquisition. De-
studies. However, studies have also found that the despite learning ASL at matched ages, the grammatical
cline in L2 grammatical performance associated with performance of the L2 learners was significantly higher
increasing age of L2 acquisition does not appear to stop and different (82%) from that of learners with no early
at any age after maturation. An important factor in the language experience (43%) on a task requiring recall of
L2 situation with respect to a possible CP for language is complex ASL sentences (Mayberry, 1993). These results
that it entails, by definition, prior acquisition of an L1 in provided the first evidence that language acquisition in
early life. Some researchers have proposed that the early life is necessary for the capacity to learn language
scope of the CP for language is restricted to L1 outcome, to develop completely, in the case where the early lan-
citing as evidence the few available case studies of social guage is spoken and the second language is signed.
isolation in early childhood (e.g., Eubank & Gregg, The goal of the present study was to replicate and
1999, among others). Cases of childhood social isolation extend our previous findings to determine whether they
are difficult to interpret given the multiple additional generalize across sensory-motor modalities and lan-
deprivations often suffered by these children (Curtiss, guages. We asked whether the acquisition of a signed
1977). Whether the postulated CP for language pri- language during early life enables the subsequent ac-
marily affects L1 outcome as compared to L2 outcome is quisition of a spoken language. We also asked the cor-
an empirical question with theoretically significant ollary question of whether a paucity of language
ramifications. We investigated this question in previous acquisition during early life attenuates the ability to
research (Mayberry, 1993, 1994). The goal of the present acquire language in later life. Such findings would mean
study was to replicate and extend these findings. that the capacity to acquire language requires early in-
One naturally occurring situation where age of L1 put from the environment to develop fully.
acquisition varies naturally and widely is the signed In order to investigate these questions, we compared
language acquisition of individuals who are born deaf the grammatical skills of hearing and deaf individuals
who learned English at similar ages but who had three hearing parents who learned and used signed language
contrasting types of language experience in early child- before the age of three. The Early-Sign Lang partici-
hood: (a) early acquisition of a spoken language from pants were later exposed to English as a second language
birth; (b) early acquisition of a signed language from when they enrolled in preschools and elementary schools
birth; and (c) little or no language acquisition during where English was the language of instruction between
early childhood. Finding superior grammatical perfor- the ages of 3 and 7 years.
mance by the early language learners compared to the The Early-Sign Lang participants were primarily
learners who experienced sparse early language would taught English through a combination of signs and
support the hypothesis that accessible language input spoken English, in addition to lipreading, reading, and
during human development is necessary for the capacity writing. The classroom language of instruction for the
to learn language to develop fully. By contrast, finding Early-Sign Lang participants was English, primarily in
similar grammatical performance by the learners with the form of simultaneously spoken English and signed
and without early language experience would provide Manually Coded English (MCE).2 These participants
counter-evidence to our hypothesis and suggest instead reported that most of their hearing teachers spoke and
that maturation alone underlies age constraints on the simultaneously signed MCE but that a few deaf teachers
capacity to learn language. Finally, finding no perfor- used ASL.
mance differences between the hearing and deaf early The length of time the Early-Sign Lang participants
language learners, whose first languages were spoken had used English was computed by subtracting age of
and signed, respectively, would mean that develop- school enrollment from chronological age. Mean length
ment of the language learning capacity is plastic with of English use was 24.5 years with a range of 14–47
respect to the sensory and motor form of the early years. The eight men and six women ranged in age from
experience and hence not a factor in the critical 17 to 52 years with a mean of 25.5 years.
period phenomenon.
2.1.3. Early spoken language
Thirteen participants had normal hearing throughout
2. Methods life and acquired a spoken language other than English
from birth; they later learned English as a second lan-
2.1. Participants guage after enrolling in a school where English was the
language of instruction (henceforth Early-Spkn Lang).
Fifty-four adults with contrasting types of language Urdu was the maternal language of eight Early-Spkn
experience in early life were recruited into the study Lang participants; French was the maternal language of
from four Canadian cities and placed in four groups. two participants; and German, Italian, and Greek were
Two groups were normally hearing and two groups were the maternal languages of three participants respec-
profoundly deaf. The groups consisted of approximately tively. The Early-Spkn Lang participants began to learn
equal numbers of men and women with the language English as a second language in schools where English
backgrounds described below and shown in Table 1. was the language of instruction between the ages of 6
and 13 years with a mean age of 9 years. After enrolling
2.1.1. Native English controls in an English school, all the Early-Spkn Lang partici-
Fourteen individuals who had normal hearing pants continued to use their maternal language at home
throughout life and were native speakers of English and socially with friends. The Early-Spkn Lang partic-
served as controls (henceforth native controls, NC). ipants had used English as an L2 for a mean of 23.46
English was their native language which they used so- years with a range of 12–50 years. The seven men and six
cially and at home, work, and school. All but one NC women ranged in age from 17 to 57 with a mean age of
participant later received second-language instruction in 32.46 years.
elementary school, French in all cases. The eight men
and six women had a mean age of 26.43 years with a
range of 16–46 years.
2
Pedagogical sign systems designed to teach English to deaf
2.1.2. Early signed language children go by a variety of names in North America, such as Manually
Fourteen participants were born profoundly deaf and Coded English (MCE) or signed English. Open class signs are
acquired ASL as a first language in early life (henceforth borrowed from a natural signed language, ASL in this case, and
Early-Sign Lang). ASL was the maternal language of 11 produced in English word order along with invented signs for portions
Early-Sign Lang participants whose deaf parents used it of English grammatical morphology. Using this communication
method, known as Total Communication or TC, the teacher speaks
with them from birth. Two participants had older deaf English and simultaneously produces the signs of MCE. The gram-
siblings who communicated with them in signed lan- matical structure of MCE is very different from that of ASL (see
guage prior to age three; one participant had normally Schick & Moeller, 1992).
Table 1
Background characteristics of the experimental groups
Early language experience n Females/ Mean age of English Mean years of English Mean chronological
Males exposure (range) use (range) age (range)
Early-Spkn English (native control) 14 6/8 Birth (0) 26.43 (16–46) 26.43 (16–46)
Early-Sign Lang (not English) 14 8/6 5.0 (4–7) 24.5 (14–47) 25.5 (17–52)
Early-Spoken Lang (not English) 13 6/7 9.00 (6–13) 23.46 (12–50) 32.46 (17–57)
No-Early Lang 13 8/6 9.40 (6–13) 26 (11–64) 35.23 (23–70)
2.1.4. No early language No-Early Lang participants and considered to be the

Thirteen participants were born profoundly deaf and age of first enrollment in a school where the language of
had normally hearing parents and siblings. Due to their instruction was English, spoken English for the former
deafness, however, the English spoken by their families group and MCE and simultaneously spoken English for
was inaccessible to them in infancy and early childhood. the latter group. Approximately half the participants in
In addition, their families neither knew nor used any each group were first exposed to accessible English be-
form of signed language with them throughout early tween the ages of 6 and 8 years (Early-Spkn Lang group,
childhood. Twelve participants first attended preschools n ¼ 8; No- Early Lang group, n ¼ 6); the remaining
for deaf children where child/teacher interaction was participants in each group were first exposed to English
restricted to speech (with no signs of any kind used) between the ages of 9 and 13 years (Early- Spkn Lang,
between the ages of 3 and 6 with a mean age of 4.1. n ¼ 6; No-Early Lang, n ¼ 7). There was no difference
However, due to a lack of speech development due to in age of first accessible English exposure between the
their profound deafness as described below (lipreading two groups (t ¼ 1:047; df ¼ 24, ns). In addition, there
alone provides insufficient linguistic input for spoken was an overlap in the age of exposure to accessible
language to develop spontaneously), they were subse- English between the above groups and the Early-Sign
quently switched to schools where sign was used, be- group; six Early-Sign participants were first exposed to
tween the ages of 6 and 13 with a mean age of 9.4. This accessible English between the ages of 4 and 7 years; the
is a common educational occurrence arising from the remaining eight participants were first exposed at age 3.
priority frequently given to speech over sign by reha-
bilitation professionals and hearing parents. One par- 2.2.2. Length of English use
ticipant attended no preschool and first enrolled in a To ensure that each participant had a baseline
school where sign was used at age 7. All these partici- amount of English experience, defined as years of use,
pants consequently had little accessible language expo- only individuals who had a minimum of 12 years of
sure until age 6 or older (henceforth No-Early Lang). continuous English use were recruited into the study.
After enrolling in a school where sign was used, the No- Overall, the participants had substantially more experi-
Early Lang participants were taught English in the same ence than the minimum requirement, as Table 1 shows.
manner as the Early-Sign Lang participants, primarily Length of English experience was computed for each
through a combination of signs and spoken English (see participant by subtracting age of first exposure to ac-
Endnote 2). They were also taught English through lip- cessible English from chronological age. Age of first
reading, reading, and writing. Several of these partici- English exposure for the native controls was considered
pants attended the same schools as the Early-Sign Lang to be at birth. Mean length of English use for the four
participants. The eight men and six women ranged in age groups, NC, Early-Sign, Early-Spkn, and No-Early
from 23 to 70 with a mean age of 35.23 years. The mean Lang, was 26.43, 24.50, 23.46, and 26.00 years, respec-
length of time the No-Early Lang participants had used tively. There were no significant differences in length of
English was 26 years with a range of 11–64 years. experience among the groups (one-way analysis of var-
iance, F ½3; 50 ¼ 0:157, ns).
2.2. Group comparability
2.2.3. Deaf group comparability
Educational background was controlled across the Although the two deaf groups had strikingly different
groups; all participants had completed high school but experiences with accessible language in early life, they
none had completed college. The following additional did not differ with respect to the age of their first pre-
factors were also controlled. school experience. The Early-Sign participants first en-
rolled in preschool programs, where MCE and
2.2.1. Age of English exposure simultaneous spoken English was the mode of child/
Age of accessible English exposure was matched teacher communication, between the ages of 3 and 5
as closely as possible for the Early-Spkn Lang and (one participant attended no preschool and entered
primary school at age 7) with a mean entry age of 4.00 sparse, perceptible and accessible language experience
years. By comparison, the No- Early Lang participants during early life. The two deaf groups, one with and one
first enrolled in preschool programs where only speech without early accessible language, did not differ with
was used between the ages of 3 and 5 (three participants respect to degree of hearing loss, non-verbal IQ, age of
attended no preschool) with a mean entry age of 4.5 preschool entry, or method of English instruction (once
years. There was no difference between the groups in age enrolled in a school where sign was used). Importantly,
of preschool entry ðt ¼ 7:11; df ¼ 22; nsÞ. the deaf group whose initial exposure to accessible lan-
Additional measures of hearing loss and non-verbal guage was delayed performed at normal levels on non-
IQ were taken on approximately half the groups who language cognitive tests. The SES of the deaf group with
were deaf, six participants each. Audiometric testing of no early language was likely higher than the group with
these 12 individuals confirmed all the participantsÕ self- early sign language. Finally, aside from the native En-
report of being profoundly deaf (mean pure-tone aver- glish controls who were first exposed to English at birth,
age, PTA, for 500, lK, and 2K Hz P90 dB for the better the three remaining groups were exposed to accessible
ear). There was no significant difference in mean PTA English at school at similar ages and had used it for a
between the Early-Sign and No-Early Lang samples, similar length of time.
94.67 and 96.20 dB, respectively ðt ¼ :55; df ¼ 10; nsÞ.
Numerous IQ studies and a meta analysis (Braden, 2.3. Experimental tasks
1992) have found that congenital deafness does not af-
fect non-verbal intelligence despite common delays in 2.3.1. Grammatical judgement
age of initial language exposure to accessible (Mayberry, A grammatical judgement task was used to assess
2000). This was confirmed by the deaf sampleÕs perfor- knowledge of selected English structures. The gram-
mance on three subtests of the Wechsler Adult Intelli- matical judgement task is commonly used to assess the
gence Scale, WAIS (Wechsler, 1981). Both groupsÕ language ability of monolingual, bilingual, and brain
performance was within the average range as compared damaged populations (Linebarger, Schwartz, & Saffran,
to the hearing population, with no significant differences 1983). English grammatical and ungrammatical stimuli
between the them (average scaled score for the hearing were presented in print, one at a time, on a computer
population is 10.00; mean scaled scores for the No-Early screen. The participant indicated with a button press
Lang and Early-Sign groups, respectively, were as whether the given stimulus was grammatical or not. The
follows: Picture Completion, 11.00 and 11.16, t ¼ computer measured both response accuracy and latency.
0:155; df ¼ 10, ns; Picture Arrangement, 10.50 and Response accuracy reflected sensitivity to syntactic
11.33, t ¼ 33; df ¼ 10, ns; Block Design, 11.33 and structures and response latency reflected the time re-
13.00, t ¼ 1:012; df ¼ 10, ns). The non-significant quired to identify these structures.
higher performance on the Block Design subtest by the
group with early signed language input likely reflects the 2.3.2. Sentence to picture matching
enhanced spatial abilities that have been found in as- In order to assess comprehension of the same selected
sociation with learning a signed language both behav- English grammatical structures, participants were given
iorally (Emmorey, Kosslyn, & Bellugi, 1993) and a second task, sentence to picture matching. Sentence to
neurocortically (Neville & Lawson, 1987). picture matching is commonly used to assess compre-
Finally, employment surveys taken in the USA have hension of selected grammatical rules in a particular
found that the social-economic status (SES) of the deaf language. The participant was shown a printed sentence
population is significantly lower than that of the hearing and asked to select the one picture from among three
population primarily due to higher levels of both un- alternatives that best depicted the stimulus sentence
employment and underemployment among adults who meaning.
are congenitally deaf. Median income of families headed
by deaf adults is approximately 70% of that of families 2.4. English grammatical structures
headed by hearing adults (Schein, 1989). Consequently,
the childhood SES of the No-Early Lang participants Five English grammatical structures were selected to
(whose 26 parents were hearing) was likely to have been represent a continuum of complexity from simple to
greater than that of the Early-Sign participants (22 of complex and were as follows: simple, dative, conjoined,
whose 28 parents were deaf). Thus, if SES were a biasing passive, and relative clause sentences. Previous research
factor in the present study, the bias would favor the No- has found that these structures represent a sequence of
Early Lang group over the Early-Sign Lang group. early to later acquired in the English acquisition of both
To summarize the experimental grouping, in addition normally hearing (Ingram, 1989) and deaf children
to the native English controls, two groups had early (Quigley & King, 1980). Because the performance of the
language experience, one in spoken language and the deaf school-aged population on these structures has
other in signed language, and the fourth group had been investigated in detail, we used them in the present
study to determine the extent to which early language grammatical judgement task and were taken from the
experience, or a lack thereof, may be a contributing Rhode Island Test of Language Structure, RITLS (En-
factor in deaf individualsÕ often poor performance on gen & Engen, 1983).
these structures. More specifically, previous research has
found that 18 year olds who were born deaf show dif- 2.5.1. Simple sentences
ficulty with syntactic structures involving multiple The simple sentences consisted of a subject and verb
clauses, as in relative clause sentences, and non-canon- followed by a prepositional phrase as in, ‘‘The girl is
ical word order, as in passive sentences (Quigley & King, playing in the water.’’ The present progressive verb tense
1980). These particular syntactic structures have not was used throughout. The grammatical stimuli were
been proposed as being especially difficult for hearing L2 made ungrammatical by exchanging the auxiliary verb
English learners. ‘‘be’’ with the auxiliary verb ‘‘have,’’ as in, ‘‘The girl
have playing in the water.’’ Previous research has found
2.4.1. Stimulus controls this to be a common error made by deaf children
To ensure that grammatical structure was the pri- (Quigley, Montanelli, & Wilbur, 1976).
mary factor that varied across the stimuli, all sentences
were to seven to ten words in length. The vocabulary 2.5.2. Dative sentences
was at an approximate grade one reading level to ensure The dative sentences consisted of a subject, verb, and
that lexical knowledge was not confounded with gram- an indirect object as in, ‘‘The boy is giving the girl a
matical processing and comprehension. For the gram- cookie.’’ The grammatical stimuli were made ungram-
matical judgement task, 24 stimuli were created for each matical with a word order violation by moving the indi-
grammatical structure. The 12 grammatical stimuli for rect object in front of the verb as in, ‘‘The boy a cookie is
each grammatical structure were made ungrammatical giving the girl.’’ Deaf children have been found to have
by the application of a single deviation that rendered it problems interpreting word order (Quigley & Power,
ungrammatical; the same deviation was then applied to 1972).
all 12 examples of the grammatical structure. When a
word was removed from a grammatical example to 2.5.3. Conjoined sentences
make it ungrammatical, a simple adjective was added to The conjoined sentences consisted of two clauses
the ungrammatical counterpart to ensure that the word joined by a temporal conjunction as in, ‘‘The girl is
length of each stimulus pair, grammatical and ungram- eating while the man is sleeping.’’ The grammatical
matical, was identical. The adjectives added were the stimuli were made ungrammatical with a word order
words ‘‘big, old, ‘‘ and ‘‘red.’’ violation by moving the conjunction to the end of the
sentence as in, ‘‘The girl is eating the man is sleeping
2.5. Stimuli while.’’ As previously explained, deaf children often
have problems interpreting English word order (Engen
Stimuli for the grammatical judgement task were 120 & Engen, 1983).
in total, 24 stimuli for each of the five English gram-
matical structures, 12 grammatical sentences and 12 2.5.4. Passive sentences
ungrammatical counterparts. The target grammatical The passive structures were non-reversible, full pas-
structures are described below and shown in Table 2. sives as in, ‘‘The boy was hit by the red ball.’’ The gram-
Stimuli for the sentence to picture matching task were matical sentences were made ungrammatical by deleting
25 in total with five example sentences of each of the five the marker ‘‘by,’’ as in ‘‘The boy was hit the big red ball.’’
target grammatical structures that appeared in the Previous research has shown the passive structure to be
Table 2
Examples of the English syntactic structures and rule violations tested
Syntactic structure Rule violation Example
Simple Auxiliary changed from ‘‘be’’ to ‘‘have’’ The girl is playing in the water
*The girl have playing in the water
Dative Indirect object placed before the verb The father is giving the girl an apple
*The father an apple is giving the girl
Conjoined clauses Conjunction placed at end of sentence The girl is eating while the man is sleeping
*The girl is eating the man is sleeping while
Non-reversible passive Deletion of passive marker ‘‘by’’ The girl was hit by the ball
*The girl was hit the ball
Subject–subject relative clause Incorrect relative clause marker The boy who is chasing the girl is happy
*The boy whose is chasing the girl is happy
very difficult for deaf children (Engen & Engen, 1983; consent were explained in English to the hearing par-
Power & Quigley, 1973). ticipants and in ASL to the deaf participants. The
grammatical judgement task was administered first fol-
2.5.5. Relative clause sentences lowed by the sentence to picture matching task.
The relative clause sentences contained a medial rela- The participant was shown how to perform the
tive clause. The subject of the main clause was modified by grammatical judgement task using three practice trials.
the relative clause and served as the subject of the relative The participant performed the practice trails by reading
clause (subject–subject relatives) as in, ‘‘The boy who is the stimulus on the computer screen and then pressing
chasing the girl is happy.’’ The grammatical sentences the green button on the game pad if he or she judged the
were made ungrammatical by substituting the incorrect stimulus to be grammatical or the red button if the
relative marker ‘‘whose’’ for ‘‘who’’ as in, ‘‘The boy whose participant judged it to be ungrammatical. ÔYesÕ re-
is chasing the girl is happy.’’ Previous research has found sponses were always made with the dominant hand; left-
relative clauses to be difficult for deaf children (Engen & handed participants made ÔyesÕ responses with the left
Engen, 1983; Quigley, Smith, & Wilbur, 1974). hand. Participants were told that the computer recorded
accuracy and latency and instructed to be careful but
2.6. Procedures not to pause unnecessarily.
After completion of the grammatical judgement task,
2.6.1. Grammatical judgement materials the sentence to picture matching task was explained in
The grammatical judgement task was presented on an English to the hearing participants and in ASL to the deaf
Apple G3 PowerBook laptop computer with an at- participants. Participants were instructed to read the
tached game pad with two buttons, green and red. The stimulus sentence at the top of the page and then to point
green button signified a ‘‘yes’’ response and the red to one of the three pictures that best depicted the sentence
button signified a ‘‘no’’ response. The grammatical meaning. Participants were given three practice trials.
judgement task was created with PowerLaboratory 1.0.3 Responses were recorded by an examiner on a score re-
experimental software (Chute & Daniel, 1996). The 120 port form. Participants had unlimited time to respond.
stimuli were presented in a fixed random order to each
participant. The computer recorded the participantÕs
response accuracy and latency. Response latency was 3. Results
recorded in milliseconds from the onset of the stimulus
to the participantÕs button press. Each stimulus presen- The groupsÕ performance on the grammatical judge-
tation was preceded by a focus signal (Ô+Õ) of 500 ms ment task was analyzed for response accuracy and la-
duration to alert the participant that the stimulus would tency and A0 as described below. For the sentence to
appear shortly on the monitor. The stimulus remained picture matching task, response accuracy was analyzed.
on the screen until the subject responded. Further analyses were conducted to compare the per-
formance of participants within the groups who were
2.6.2. Sentence to picture matching materials first exposed to accessible English at earlier (6–8 years)
The sentence to picture matching task was presented versus older (9–13 years) ages.
on paper in a binder format. The printed stimulus sen-
tence was centred at the top of the page and the three 3.1. Grammatical judgement task
alternative pictures were given below on the bottom.
The stimulus sentences and pictures were taken from the 3.1.1. Response accuracy
RITLS (Engen & Engen, 1983). The pictures showed The participantsÕ response accuracy on the gram-
three possible grammatical interpretations of the stim- matical judgement task was analyzed with two, 4 5 2
ulus sentence rather than lexical or morphological con- repeated measures analyses of variance for subjects and
trasts. For example, the stimulus sentence, ‘‘The woman items. These analyses revealed whether the effects held
who is holding the baby has a hat on,’’ was accompanied across both the participants and the stimuli and further
by three pictures each depicting two women with one determined whether grammaticality was a performance
holding a baby; in one foil the woman not holding the factor. The between-subjects factor was early language
baby wears a hat and in the other foil there is no hat. experience with four levels of group, NC, Early-Sign,
There were 25 stimuli, five examples of each of the five Early-Spkn, and No-Early Lang. The within-subjects
selected grammatical structures. The 25 stimuli were factors were: (1) syntactic structure with five levels of
presented in a fixed random order. type (simple, dative, conjoined, passive, and relative
clause) and (2) grammaticality with two levels, gram-
2.6.3. Participant testing matical and ungrammatical.
Each participant was tested individually in a single The results showed a significant main effect for early
session. The nature of the tasks, stimuli, and informed language experience (FSubjects ½3; 50 ¼ 18:58; p < :0001;
FItems ½3; 165 ¼ 405:88; p < :0001). Mean error rate was 4yð1 xÞ (Linebarger et al., 1983). A0 is a form of signal
5% for the NC, 14% for the Early-Sign, 14% for the detection that factors guessing behavior into performance
Early-Spkn, and 34% for the No-Early Lang group. by comparing the ratio of hits (x ¼ correct judgements of
There was a significant main effect for grammatical ungrammatical stimuli) and false alarms (y ¼ incorrect
structure that interacted in a two-way interaction with judgements of grammatical stimuli) as a function of
early language experience (FSubjects ½12; 200 ¼ 2:910; chance. A0 scores range from 0.5 suggesting no rule sen-
p < :001; FItems ½12; 165 ¼ 8:127; p < :001) and in a sitivity to 1.0 showing high rule sensitivity.
three-way interaction with grammaticality ðFSubjects We analyzed the A0 scores with a 4 5 analysis of
½4; 200 ¼ 14:010; p < :001; FItems ½4; 55 ¼ 7:826; p < :001Þ. variance for subjects only. The between-subjects factor
There was no main effect for grammaticality, however. was early language experience with four levels of group.
The nature of the three-way interaction was such that The within-subjects factor was grammatical structure
the performance of the No-Early Lang group on the with five levels of rule type. Grammaticality was not a
ungrammatical examples of the simple, passive, and factor because it was incorporated into the A formula in
relative clause structures was significantly lower than the form of ÔhitsÕ and ÔmissesÕ as explained above. The
their performance on the grammatical counterparts of results showed that guessing behavior was not a biasing
these structures. In addition, their accuracy on the un- factor in the relative performance patterns of the groups
grammatical examples of the simple, passive, and rela- across the syntactic structures. First, there was a sig-
tive clause structures was significantly less than that of nificant main effect for early language experience
the other three groups. The performance of the early (F ½3; 50 ¼ 18:881; p < :001). The mean A0 scores were
language groups, NC, Early-Sign Lang, and Early-Spkn .93 for NC, .83 for the Early-Sign Lang, .82 for the
Lang, was not significantly different from one another Early-Spkn Lang, and .68 for the No-Early Lang group.
independent of grammatical structure or grammatical- There was also a significant main effect for type of
ity, with one exception. The Early-Spkn Lang group was grammatical structure (F ½4; 200 ¼ 18:945; p < :001)
significantly less accurate on the grammatical examples that interacted with early language experience
of the dative structure compared with the performance (F ½12; 200 ¼ 2:897; p < :001), as shown in Fig. 1. The
of the NC and Early-Sign Lang groups on these stimuli nature of the interaction was such that the No-Early
(Student–Newmann–Keuls, p < :05 for each compari- Lang group obtained A0 scores that were significantly
son with a harmonic mean for group n). lower on the simple, passive, and relative clause struc-
tures than those of the three groups who had early
3.1.2. Response A0 language experience. In addition, the No-Early Lang
To determine the extent to which the above results were group obtained A0 scores on the conjoined structure that
biased by guessing behavior, we further analyzed the were significantly lower than those of the NC but not the
groupsÕ performance by calculating an A0 score for each Early-Sign Lang and Early Spkn-Lang groups. Finally,
participant for each of the five grammatical structures. the Early-Spkn Lang group obtained A0 scores on the
The formula we used was: 0:5 þ ½ðy þ xÞð1 þ y xÞ= dative structure that were significantly lower than those
Fig. 1. Performance accuracy on a grammatical judgement task expressed as mean A0 score (that takes guessing into account) as a function of English
syntactic structure and early language experience, native English control (NC), early sign language learners, early spoken language learners, and no
early language learners. Chance performance would be at 0.50 as indicated by the minimum on the Y-axis.
of the NC group but not the Early-Sign Lang and or interaction effects for grammaticality in either the
No-Early Lang groups (Student–Newmann–Keuls, subject or item analyses.
p < :05 for each comparison with a harmonic mean The effect of early language experience on grammat-
for group n). ical judgement response latency was such that each
groupÕs response latency was significantly different from
3.1.3. Response latency the otherÕs, as shown in Table 3. The NC group re-
Response latencies for correct grammatical judge- sponded more quickly than the three other groups. The
ments were analyzed with two, 4 5 2 repeated No-Early Lang group responded more slowly than the
measures analyses of variance for subjects and items in three other groups. Of the two early language groups,
order to determine whether grammatical processing rate the Early-Sign Lang participants responded more
is also affected by the onset of accessible language quickly than did the Early-Spkn Lang participants
experience. The between-subjects factor was early lan- (Student–Newmann–Keuls, p < :05 for each compari-
guage experience with four levels of group. The within- son with a harmonic mean for group n).
subjects factors were: (1) grammatical structure, with
five levels of type and (2) grammaticality with two levels, 3.2. Sentence to picture matching task
grammatical and ungrammatical.
The results showed a significant main effect for early The participantsÕ scores on the sentence to picture
language experience (FSubjects ½3; 50 ¼ 13:622; p < :001; matching task were analyzed with two 4 5 repeated
FItems ½3; 165 ¼ 134:671; p < :001). There was also a measures analyses of variance for subjects and items.
main effect for grammatical structure that was signifi- The between-subjects factor was early language experi-
cant in the subject analysis (FSubjects ½4; 200 ¼ 5:086; ence with four levels of group; the within-subjects factor
p < :001) but not in the item analysis. There were no was grammatical structure with five levels of rule type.
significant interactions between early language experi- The results showed a significant main effect for early
ence and grammatical structure and no significant main language experience ðFSubjects ½3; 30 ¼ 24:06; p < :0001;
FItems ½3; 60 ¼ 26:225; p < :001Þ. Mean comprehension
accuracy was 97% for the NC, 94% for the Early Sign
Table 3 Lang group, 95% for the Early-Spkn Lang group, and
Grammatical judgement response latency in milliseconds as a function 74% for the No-Early Lang group. The results also
of early language experience (for correct responses only)
showed a significant main effect for grammatical struc-
Group Mean SD ture ðFSubjects ½4; 200 ¼ 20:39; p < :0001; FItems ½4; 20
Native control 3880 1570 ¼ 5:308; p < :000lÞ that further interacted with early
Early-Sign Lang 4778 2171 language experience ðFSubjects ½12; 200 ¼ 11:88; p <
Early-Spkn Lang 6325 4594 :0001; FItems ½12; 60 ¼ 4:688; p < :0001Þ, as Fig. 2
No-Early Lang 8492 4377
shows. The nature of the interaction was such that the
Fig. 2. Mean performance accuracy on a sentence to picture matching task expressed as proportion correct as a function of English syntactic
structure and early language experience, native English control (NC), early sign language learners, early spoken language learners, and no early
language learners. Chance performance would be 0.33.
groupsÕ comprehension accuracy did not differ signifi- exposure occurred at younger ages (6–8 years) or older
cantly across the simple, dative and conjoined struc- (9–13 years) ages on the simple (F ½4; 28 ¼ 7:02;
tures. However, the No-Early Lang group was p < :001), passive (F ½4; 28 ¼ 9:21; p < :001) and relative
significantly less accurate than the three other groups on clause (F ½4; 28 ¼ 5:78; p < :001) structures (FischerÕs
the passive and relative clause structures, performing at PLSD for each comparison, p < :01). In addition, there
near chance levels (Student–Newmann–Keuls, p < :05 were no performance differences among the younger and
for each comparison with a harmonic mean for group older age subgroups who experienced language in early
n). life, as shown in Table 4.
3.3. Subgroup analyses 3.3.2. Grammatical judgement response latency

We analyzed the mean grammatical judgement re-
To determine whether non-verbal cognitive skills sponse latency with one-way analysis of variance for
were a biasing factor in the results for the groups who the five subgroups without the factors of syntactic
were born deaf, we compared the performance of the structure and grammaticality because the main analysis
participants in each group who received non-verbal IQ showed no significant effects for these factors. The
testing with those who did not. The results showed no subgroups who had early language responded signifi-
difference within groups related to non-verbal IQ per- cantly more quickly than the subgroups with no early
formance (Early-Sign, t ¼ 1:144; df ¼ 12; ns; No-Early language, except for the one subgroup who first
Lang, t ¼ 1:78; df ¼ 11; ns). learned English at older (9–13 years) ages (F ½4:28 ¼
To determine next whether age of exposure to ac- 3:66; p < :02; FischerÕs PLSD for each comparison,
cessible English within the groups may have biased the p < :05). There were no differences within each group,
pattern of results, we sorted the Early-Spkn Lang and Early-Spkn Lang and No Early Lang, as a function of
No-Early Lang groups into subgroups of younger (6-8 subgrouping.
years) versus older (9–13 years) ages of exposure to ac-
cessible English and reanalyzed the grammatical judge- 3.3.3. Grammatical comprehension
ment and comprehension data. We included the Finally, we analyzed the sentence to picture matching
subgroup of Early-Sign Lang participants whose age of performance of the five subgroups with one-way anal-
first exposure to accessible English (4–7 years) was yses of variance separately for each syntactic structure.
comparable, although not identical, to the younger age All the subgroups who had early language experience
subgrouping of the other two groups. significantly outperformed those who did not, regard-
less of whether their first exposure to English was at
3.3.1. Grammatical judgement accuracy younger (6–8 years) or older (9–13 years) ages, on the
We analyzed the adjusted grammatical judgement passive (F ½4; 28 ¼ 26:29; p < :001) and relative clause
performance (A0 scores described above) with one-way (F ½4; 28 ¼ 19:64; p < :001) structures (FischerÕs PLSD
analyses of variance separately for each syntactic for each comparison, p < :01). Finally, there were no
structure. All subgroups who had early language expe- performance differences among the younger and older
rience significantly outperformed those who did not, subgroups who experienced language in early life, as
regardless of whether their first accessible English shown in Table 5.
Table 4
Adjusted mean grammatical judgement accuracy of the Early-Sign Lang, Early-Spkn Lang and No-Early Lang participants subgrouped by age of
exposure to accessible English
Subgroup Syntactic structure
Simple, mean Dative, mean Conjoined, mean Passive, mean Relative clause,
(SD) (SD) (SD) (SD) mean (SD)
Early-Sign Lang
4–7 years (n ¼ 7) 0.87 (0.14) 0.79 (0.15) 0.88 (0.09) 0.95 (0.05) 0.79 (0.17)
Early-Spkn Lang
6–8 years (n ¼ 7) 0.95 (0.06) 0.68 (0.10) 0.85 (0.06) 0.93 (0.11) 0.83 (0.13)
9–13 years (n ¼ 6) 0.88 (0.17) 0.72 (0.20) 0.94 (0.05) 0.84 (0.16) 0.86 (0.21)
No-Early Lang
6–8 years (n ¼ 8) 0.69 (0.05) 0.67 (0.10) 0.74 (0.17) 0.69 (0.15) 0.57 (0.13)
9–13 years (n ¼ 5) 0.63 (0.04) 0.66 (0.15) 0.75 (0.09) 0.63 (0.04) 0.54 (0.04)
*
p < :01.
Table 5
Mean grammatical comprehension accuracy of the Early-Sign and Spkn-Lang and No-Early Lang participants subgrouped by age of accessible
English exposure
Subgroup Syntactic structure
Simple, mean Dative, mean Conjoined, mean Passive, mean Relative clause,
(SD) (SD) (SD) (SD) mean (SD)
Early-Sign Lang
4–7 years (n ¼ 7) 1.00 (0) 0.96 (0.08) 1.00 (0) 1.00 (0) 0.90 (0.16)
Early-Spkn Lang
6–8 years (n ¼ 7) 1.00 (0) 1.00 (0) 0.94 (0.09) 0.96 (0.07) 1.00 (0)
9–13 years (n ¼ 6) 0.92 (0.10) 1.00 (0) 0.96 (0.08) 0.80 (0.44) 0.84 (0.16)
No-Early Lang
6–8 years (n ¼ 8) 1.00 (0) 0.96 (0.07) 0.82 (0.18) 0.62 (0.33) 0.42 (0.08)
9–13 years (n ¼ 5) 0.96 (0.08) 1.00 (0) 0.88 (0.10) 0.08 (0.10) 0.36 (0.08)
*
p < :01.
4. Discussion 1993). Early experience with a spoken language led to

near-native performance on a task involving complex
The results address a fundamental question about the ASL structures whereas a lack of such experience did
nature of language acquisition and the postulated criti- not. Together the results of our previous and present
cal period for language. Does age equally constrain the studies suggest that language experience during human
outcome of all language acquisition regardless of whe- development dramatically alters the capacity to learn
ther the language is a first or second one? To answer the language throughout life and that these effects are su-
question, we investigated the English syntactic abilities pramodal with respect to both the first and second
of deaf and hearing adults who either did or did not language (Mayberry, Lock, & Kazmi, 2002).
have linguistic experience (spoken or signed) during The contrastive grammatical judgement and com-
early childhood with two tasks, timed grammatical prehension patterns of the groups with and without
judgement and untimed sentence to picture matching. early language provide clues about the nature of this
On the grammatical judgement task, adults who had critical period phenomenon. First, the syntactic pro-
early language experience performed at near-native cessing patterns of the two groups who had early lan-
levels on a subsequently learned language, regardless of guage experience were remarkably similar across tasks.
whether the early language was spoken or signed. By The similarities were striking given that one group first
contrast, adults who had little or no early language ex- learned a signed language and the other a variety of
perience performed poorly on several syntactic struc- spoken languages and that one group was born pro-
tures, namely, simple, passive and relative clause foundly deaf and the other normally hearing. None-
sentences. On the sentence to picture matching task, theless, the grammatical judgement performance of both
adults who had early language again performed at near- groups with early language was similar to that of the
native levels regardless of whether their early language native controls with one exception. Both groups were
was signed or spoken. Again by contrast, adults who less accurate on the dative sentences compared to the
had no early language experience performed at near- native controls. In creating these stimuli, we made un-
chance levels on the complex structures of passive and grammatical stimuli that would have been grammatical
relative clause sentences. These findings held when the if translated into some of these participantsÕ native
groups were subdivided into earlier ages of exposure to languages, spoken or signed, by placing the indirect
accessible English (6–8 years) versus older ages (9–13 object in front of the verb, as in, ‘‘The boy a cookie is
years). Finally, the age of first-language experience also giving the girl.’’ Many Early-Spkn and Early-Sign Lang
affected grammatical judgement response latency. participants accepted these stimuli as grammatical. This
Adults who had early language experience and began to suggests that early language acquisition leads to the
learn English before age 9 recognized English syntactic development of both general and specific syntactic rep-
structures more quickly than those who had no early resentations that organize and subserve subsequent
language, although more slowly than native English language learning. By syntactic representations we mean
learners, independent of syntactic structure and re- mental representations of syntactic categories organized
gardless of whether the early language experience was in a hierarchical fashion (Jackendoff, 1993).
signed or spoken. The early-language groups showed high levels of ac-
The results replicate and extend our previous findings curacy on both tasks across the syntactic structures in-
both cross-modally and cross-linguistically (Mayberry, dicating that their syntactic representations for English
grammar were robust. The additional finding that the produces decrementally poorer language performance
early language groups were somewhat slower in access- (Boudreault, 1999; Mayberry, 1993, 1994; Mayberry &
ing these syntactic representations in comparison to the Eichen, 1991; Mayberry & Fischer, 1989). Current re-
native controls suggests either that their syntactic rep- search in our laboratory finds that even longer delays in
resentations of English grammar are not identical to accessible first-language exposure, i.e., age 13 or older,
those of native English learners or, alternatively, that produces even lower levels of grammatical ability than
their processing of English grammar is not as fully those reported here (Gates, 2002). If the present and
automatized as that of native learners. The question is previous effects were simply pedagogic in origin, they
whether these effects arise from syntactic competence or would not be expected to show such consistent, lifelong,
performance factors. However, this somewhat slower and linear decrements in relation to the duration of
rate of grammatical processing cannot be due to the fact language deprivation in early life.
that the early language groups knew and used two lan- The low levels of grammatical judgement accuracy
guages because all but one of the native controls also shown by the group with no early language suggests that
knew more than one language. Rather, slower, but ac- they have incomplete or missing representations of En-
curate, grammatical processing latencies may reflect glish syntax for both simple and complex structures.
successive language acquisition in childhood, a second Unlike the groups who had early language experience,
language acquired after a native one. Response latency they did not perform the same across the tasks. Only
decrements of a similar magnitude for grammatical linguistic information was available for the grammatical
judgements have been reported by White and Genesee judgement task but extra-linguistic information was
(1996) as a function of second-language proficiency. additionally available for the sentence to picture
The grammatical processing patterns of the group matching task. Exact representations of the English
with scant early language experience were strikingly auxiliary system, word order, and subordination, for
different from those of the groups who had early lan- example, were required to perform well on the gram-
guage. These differences were clearly due to a paucity of matical judgement task. This may not have been the case
accessible and detailed linguistic input in early life and for the picture to sentence matching task where correct
not deafness. This is shown by the high performance guesses could have been made via an alternative strategy
level of the early language group who was also born such as identifying the meaning of some words without
profoundly deaf but who experienced accessible lan- the organizing benefit of syntax. A strategy based on
guage early; their performance was at near-native levels, isolated word meaning coupled with guessing would
like their hearing peers who also experienced language likely have led to wrong responses for the complex
early. It is important to note that this was also true of passive and relative clause sentences on the sentence to
the deaf participants in the Early-Sign Lang group who picture matching task, however. More than two agents
were first exposed to signed language by age 3. This were illustrated in the alternative pictures for the relative
suggests that early childhood is a period of robust sen- clause sentences, and multiple actions/relations were il-
sitivity to accessible linguistic input. The grammatical lustrated for the passive sentences. The only route to
processing deficits associated with a lack of early lan- correct responses for such sentences was through iden-
guage were not due to factors such as non-verbal IQ, tification of syntax. In addition, the finding that the
SES, access to preschool education, method of English mean response latencies of the group with no early
instruction in schools where sign was used, or highest language was slower than that of the groups who had
level of education attained because these factors did not early language and learned English by age 9, indepen-
differ between the groups who were born deaf. The dent of syntactic structure, coupled with their low ac-
primary contrast was that one group experienced accuracy levels, provides support for this interpretation.
cessible language from birth, or shortly thereafter, while A lack of accessible language experience in early life
the other did not. The finding that the group with no appears to impede development of syntactic represen-
early language was unable to acquire English grammar tations in any subsequently learned language, indepen-
well after many years of daily usage means that these dent of sensory-motor modality. This is demonstrated
effects are long lasting. by the present results, where English was the target
The question arises as to whether these effects are language and by our previous findings where ASL was
pedagogic in nature and not due a critical period for the target language. Children who are born deaf and
language, that the child who begins his or her educa- have no accessible early language have been found to
tional experience with little or no language is unable to show gesture ordering patterns that are similar across
comprehend the language of instruction. This is most cultures and suggestive of syntactic categories organized
certainly the case for all children learning languages in hierarchically (Goldin-Meadow & Mylander, 1998).
immersion situations but unlikely the cause of the However, this expressive gesture, known as homesign,
present findings. In previous research we discovered that does not appear to provide sufficient linguistic experi-
increasing delay in accessible first-language exposure ence for the capacity to learn language to develop
completely. The present participants who were born growth that are also apparent at a behavioral level in at
deaf and had no early language presumably used gesture least two ways. First, animals raised in complex envi-
as young children to communicate with their hearing ronments show increased dendritic branching during the
families yet, as described above, as adults they show exuberant phase of neural development in comparison
attenuated levels of syntactic development in any lan- to those raised in impoverished environments (Kolb,
guage, spoken or signed. More research is required to Forgie, Gibb, Gorny, & Rowntree, 1998). Animals
tease apart the complex relations among the onset of raised in complex environments also maintain more
language acquisition, the nature of the linguistic input, synaptic connections during the pruning phase of neu-
and the development of syntactic representation. Whe- rophysiological development in contrast to those raised
ther early language experience affects the development in impoverished environments (Greenough & Black,
of other linguistic representations, including phonolog- 1992). Thus, complex environmental stimulation during
ical and lexical ones, is an important question we are both the phases of neuronal exuberant growth and
currently investigating in our laboratory. pruning (Huttenlocher, 1990) produces more complex
Much previous research has found that the acquisi- neurophysiological development and concomitantly
tion of English grammar is problematic for students more complex behavior in comparison to a lack of such
who are born severely and profoundly deaf (Engen & stimulation from the environment. If the same kind of
Engen, 1983; Quigley & King, 1980, among many oth- reciprocal interaction occurs in early human life between
ers). For example, Quigley et al. (1974) reported that 18 linguistic stimulation from the environment and devel-
year olds who were deaf performed with 56% accuracy opment of the neurocortical systems that subserve lan-
on relative pronouns and 59% on relative clauses on a guage, this would mean that the common interpretation
task requiring detection of grammatical errors in written of the critical period for language needs to be refor-
sentences. This syntactic performance is remarkably mulated. Instead of being a phenomenon of diminishing
consistent with the present findings, where mean accu- ability to learn language caused by increasing brain
racy on the grammatical judgement task for the group growth, the critical period for language would instead be
with no early language was 54% for relative clause a time-delimited window in early life where the degree
sentences and 45% on the comprehension task. These and complexity of neurocortical development underly-
similarities are not a coincidence but likely reflect the ing the language system is governed, in part, by lin-
limited access to language in early life commonly expe- guistic stimulation from the environment which together
rienced by children who are born deaf. Evidence is ac- with neurocortical development creates the capacity to
cumulating that the deleterious linguistic effects arising learn language. The present findings better fit the latter
from a lack of early language is a contributing factor to explanation than the former.
the low levels of reading achievement endemic to the To summarize, we have found evidence that age
deaf population as well (Chamberlain, 2002; Chamber- constraints on the outcome of first-language acquisition
lain & Mayberry, 2000). are quite different from those for second languages.
Our present and previous findings are also consistent These findings illuminate the postulated critical period
with what is known about the dramatic effects of early for language by suggesting that early language experi-
experience on the development of other biological sys- ence helps create the ability to learn language through-
tems (Changeux, 1985). Language may be a genetically out life, independent of sensory-motor modality.
specified ability but our previous and present results Conversely, a lack of language experience in early life
suggest that the development of language capacity may seriously compromises development of the ability to
be an epigenetic process whereby environmental expe- learn any language throughout life. These findings mean
rience during early life drives and organizes the growth that timely first-language acquisition is necessary, but
of this complex behavioral and neurocortical system. not sufficient, for the successful outcome of second
For example, complete development of the neurocorti- language learning.
cal architecture of the visual system crucially depends on
the onset and type of visual stimulation experienced
during early life. Visual deprivation in early childhood
profoundly alters lifelong visual abilities in animals References
(Wiesel, 1982) and humans (Goldberg, Maurer, Lewis,
& Brent, 2001). Our findings suggest that the human Baynton, D. C. (1996). Forbidden signs: American culture and the
language system may develop in much the same fashion, campaign against sign language. Chicago: The University of
as first proposed by Snow (1987). Chicago Press.
Experiments investigating the effects of enriched Bialystok, E., & Hakuta, K. (1999). Confounded age: Linguistic and
cognitive factors in age differences for second language acquisition.
versus impoverished environments on the brain devel- In D. Birdsong (Ed.), Second language acquisition and the critical
opment of rats during early life elucidate the complex period hypothesis (pp. 161–181). Mahwah, NJ: Lawrence Erlbaum
interactions between early experience and neurocortical Associates.
Birdsong, D. (1992). Ultimate attainment in second language acqui- Ingram, D. (1989). First language acquisition. Cambridge, MA:
sition. Language, 64, 706–755. Cambridge University Press.
Birdsong, D. (1999). Whys and why nots of the critical period Jackendoff, R. (1993). Languages of the mind: Essays on mental
hypothesis for second language acquisition. In D. Birdsong (Ed.), representation. Cambridge: MIT Press.
Second language acquisition and the critical period hypothesis (pp. 1– Johnson, J., & Newport, E. (1989). Critical period effects in second
22). Mahwah, NJ: Lawrence Erlbaum Associates. language learning: The influence of maturational state on the
Birdsong, D., & Molis, M. (2001). On the evidence for maturational acquisition of English as a second language. Cognitive Psychology,
constraints in second language acquisition. Journal of Memory and 21, 60–69.
Language, 44, 235–249. Johnson, J., & Newport, E. (1991). Critical period effects on universal
Boudreault, P. (1999). Grammatical processing in American sign properties of language: The status of subjacency in the acquisition
language: Effects of age of acquisition and syntactic complexity. of second language. Cognition, 39, 215–258.
Unpublished Masters Thesis, McGill University, Montreal. Johnson, J. S. (1992). Critical period effects in second language
Braden, J. P. (1992). Intellectual assessment of deaf and hard-of- acquisition: The effect of written versus auditory materials in the
hearing people: A quantitative and qualitative research synthesis. assessment of grammatical competence. Language Learning, 42(2),
School Psychology, 21, 82–94. 217–248.
Chamberlain, C. (2002). Reading skills of deaf adults who sign: Good Klima, E. S., & Bellugi, U. (1979). The signs of language. Cambridge:
and poor readers compared. Unpublished Doctoral Dissertation, Harvard University Press.
McGill University, Montreal. Kolb, B., Forgie, M., Gibb, R., Gorny, G., & Rowntree, S. (1998).
Chamberlain, C., & Mayberry, R. I. (2000). Theorizing about the Age, experience and the changing brain. Neuroscience & Biobehav-
relation of ASL to reading. In C. Chamberlain, J. Morford, & R. I. ioral Reviews, 22, 143–159.
Mayberry (Eds.), Language acquisition by eye. Mahwah, NJ: Lenneberg, E. (1967). Biological foundations of language. New York,
Lawrence Erlbaum Associates. NY: John Wiley.
Chamberlain, C., Morford, J. P., & Mayberry, R. I. (Eds.). (2000). Lillo-Martin, D. (1999). Modality effects and modularity in language
Language acquisition by eye. Mahwah, NJ: Lawrence Erlbaum acquisition: The acquisition of American sign language. In W. C.
Associates. Ritchie, & T. K. Bhatia (Eds.), Handbook of child language
Changeux, J. P. (1985). Neuronal man. New York: Pantheon. acquisition (531, p. 567). New York: Academic Press.
Chute, D. L. & Daniel, R. S. (1996). B/C PowerLaboratory (Version Lederberg, A., & Everhart, V. S. (1998). Communication between deaf
1.0.3). Pacific Grove, CA: Brooks/Cole Publishing Company. children and their hearing mothers: The role of language, gesture,
Colombo, J. (1982). The critical period concept: Research, methodol- and vocalizations. Journal of Speech, Language, and Hearing
ogy, and theoretical issues. Psychological Bulletin, 91(2), 260–275. Research, 41, 887–899.
Curtiss, S. R. (1977). Genie: A linguistic study of a modern day ‘‘wild Linebarger, M., Schwartz, M., & Saffran, E. (1983). Sensitivity to
child’’. New York, NY: Academic Press. grammatical structure in so-called agrammatic aphasics. Cognition,
Emmorey, K. (2002). Language, cognition, and the brain: Insights from 13, 361–392.
sign language research. Mahwah, NJ: Lawrence Erlbaum. Mayberry, R. I. (1993). First-Language acquisition after childhood
Emmorey, K., Bellugi, U., Friederici, A., & Horn, P. (1995). Effects of differs from second-language acquisition: The case of American
age of acquisition on grammatical sensitivity: Evidence from on- Sign Language. Journal of Speech and Hearing Research, 36, 51–
line and off- line tasks. Applied Psycholinguistics, 16, 1–23. 68.
Emmorey, K., Kosslyn, S. M., & Bellugi, U. (1993). Visual imagery Mayberry, R. I. (1994). The importance of childhood to language
and visual-spatial language: Enhanced imagery abilities in deaf and acquisition: Evidence from American Sign Language. In J. C.
hearing ASL signers. Cognition, 46, 139–181. Goodman, & H. C. Nusbaum (Eds.), The development of speech
Engen, E., & Engen, T. (1983). Rhode Island Test of Language perception (pp. 57–90). Cambridge, MA: MIT Press.
Structure. Baltimore: University Park Press. Mayberry, R. I. (2002). Cognitive development of deaf children: The
Eubank, L., & Gregg, K. R. (1999). Critical periods and (second) interface of language and perception in neuropsychology. In S. J.
language acquisition: Divide et impera. In D. Birdsong (Ed.), Segalowitz, & I. Rapin (Eds.), Handbook of neuropsychology, Part
Second language acquisition and the critical period hypothesis (pp. II (Vol. 8, 2nd ed., pp. 71–107). Amsterdam: Elsevier.
65–99). Mahwah, NJ: Lawrence Erlbaum Associates. Mayberry, R. I., & Eichen, E. (1991). The long-lasting advantage of
Flege, J. E. (1999). Age of learning and second language speech. In D. learning sign language in childhood: Another look at the critical
Birdsong (Ed.), Second language acquisition and the critical period period for language acquisition. Journal of Memory and Language,
hypothesis (pp. 101–132). Mahwah, NJ: Lawrence Erlbaum Ass- 30, 486–512.
ocaites. Mayberry, R. I., & Fischer, S. D. (1989). Looking through phonolog-
Flege, J. E., Yeni-Komshian, G. H., & Liu, S. (1999). Age constraints ical shape to lexical meaning: The bottleneck of non-native sign
on second-language acquisition. Journal of Memory and Language, language processing. Memory & Cognition, 17, 740–754.
41, 78–104. Mayberry, R. I., Lock, E., & Kazmi, H. (2002). Linguistic ability and
Gates, A. (2002). The effects of delayed first-language exposure on early language exposure. Nature, 417, 38.
language acquisition: A case study. Unpublished Masters Thesis, Morford, J. P., & Kegl, J. A. (2000). Gestural precursors to linguistic
McGill University, Montreal. constructs: How input shapes the form of language. In D. McNeill
Goldberg, M. C., Maurer, D., Lewis, T. L., & Brent, H. P. (2001). The (Ed.), Language and gesture (pp. 358–387). Cambridge, UK:
influence of binocular visual deprivation on the development of Cambridge University Press.
visual–spatial attention. Developmental Neuropsychology, 19, 55–81. Morford, J. P., & Mayberry, R. I. (2000). A reexamination of ‘‘early
Goldin-Meadow, S., & Mylander, C. (1998). Spontaneous sign systems exposure’’ and its implications for language acquisition by eye. In C.
created by deaf children in two cultures. Nature, 391, 279–281. Chamberlain, J. Morford, & R. Mayberry (Eds.), Language acqui-
Greenough, W. T., & Black, J. E. (1992). Induction of brain structure sition by eye (pp. 111–128). Mahwah, NJ: Lawrence Erlbaum and
by experience: Substrates for cognitive development. In M. R. Associates.
Gunna, & C. A. Nelson (Eds.), Developmental behavioral neurosci- Neville, H. J., & Lawson, D. (1987). Attention to central and
ence (24, pp. 155–200). Hillsdale, NJ: Lawrence Erlbaum. peripheral visual space in a movement detection task: An event-
Huttenlocher, P. R. (1990). Morphometric study of human cerebral related potential and behavioral study. II. Congenitally deaf adults.
cortex development. Neuropsychologia, 517–527. Brain Research, 405, 268–283.
Newport, E. (1990). Maturational constraints on language learning. Quigley, S. P., & Power, D. (1972). The development of syntactic
Cognitive Science, 14, 11–28. structure in the language of deaf children. Urbana, IL: Institute for
Oyama, S. (1978). The sensitive period and comprehension of speech. Research on Exceptional Children.
Working Papers on Bilingualism, 16, 1–17. Schein, J. D. (1989). At home among strangers: Exploring the deaf
Patkowski, M. (1980). The sensitive period for the acquisition of community in the United States. Washington, DC: Gallaudet
syntax in a second language. Language Learning, 30, 449– University Press.
472. Schein, J. D., & Delk, M. T. (1974). The deaf population of the United
Penfield, W., & Roberts, L. (1959). Speech and brain mechanisms. New States. Silver Spring, MD: National Association of the Deaf.
York: Atheneum. Schick, B., & Moeller, M. P. (1992). What is learnable in manually
Petitto, L. A., & Marentette, P. (1991). Babbling in the manual coded English sign systems? Applied Psycholinguistics, 3, 313–340.
mode: Evidence for the ontogeny of language. Science, 251, Senghas, A., & Coppola, M. (2001). Children creating language: How
1483–1496. Nicaraguan Sign Language acquired a spatial grammar. Psycho-
Power, D., & Quigley, S. (1973). Deaf childrenÕs acquisition of the logical Science, 12, 323–328.
passive voice. Journal of Speech and Hearing Research, 16, 5– Snow, C. (1987). Relevance of the notion of a critical period to
11. language acquisition. In M. H. Bornstein (Ed.), Sensitive periods in
Quigley, S., & King, C. (1980). Syntactic performance of hearing development: Interdisciplinary perspectives (pp. 183–209). Hillsdale,
impaired and normal hearing individuals. Applied Psycholinguis- NJ: Lawrence Erlbaum.
tics, 1, 329–356. Wechsler, D. (1981). Wechsler Intelligence Scale—Revised. New York:
Quigley, S., Montanelli, D. S., & Wilbur, R. B. (1976). Some aspects of Psychological Corporation.
the verb system in the language of deaf students. Journal of Speech White, L., & Genesee, F. (1996). How native is near-native. The issue
and Hearing Research, 19, 536–550. of ultimate attainment in adult second language acquisition.
Quigley, S., Smith, N., & Wilbur, R. (1974). Comprehension of Second Language Research, 12, 233–265.
relativized sentences by deaf students. Journal of Speech and Wiesel, T. N. (1982). Postnatal development of the visual cortex and
Hearing Research, 17, 325–341. the influence of the environment. Nature, 299, 583–591.

Mayberry Lock 2003 PDF

Uploaded by

Copyright:

Available Formats

Mayberry Lock 2003 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mayberry Lock 2003 PDF

Uploaded by

Copyright:

Available Formats

Brain and Language 87 (2003) 369–384

Age constraints on ﬁrst versus second language acquisition:

1. Introduction (1959) ﬁrst proposed that language acquisition was re-

2.1.4. No early language No-Early Lang participants and considered to be the

3.3. Subgroup analyses 3.3.2. Grammatical judgement response latency

4. Discussion 1993). Early experience with a spoken language led to

You might also like