Content uploaded by Eve Higby
Author content
All content in this area was uploaded by Eve Higby on Sep 19, 2015
Content may be subject to copyright.
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
645
[645--664] 11.4.2015 6:48PM
29
Losing a first language
to a second language
Eve Higby and Loraine Obler
When you’re a teenager you have so many other concerns,
and, [my first language] just sort of slipped away
and I didn’t realize what I had lost until it was gone.
(Greta, who immigrated to Canada from the Netherlands as a child
(Kouritzin, 1999: 171))
29.1 Introduction
The loss of one’s native language seems more distressing than the loss of a
later learned language. The native language is inextricably linked to one’s
history and sense of self. Even more than that, it is most often tied to
associations with family, tradition, culture, and heritage, making it an
integral part of one’s identity. Indeed, the shift in language dominance to
a later learned language has sometimes been described by people who have
undergone it as a shift in personality or self-identification (e.g., Kouritzin,
1999). Sometimes the suppression of the nativelanguage is intentional, as in
the case of immigrants who want to assimilate as quickly as possible into
the new society (e.g., Isurin, 2005) or in the case of strong societal pressures
against speaking one’s native language (Schmid, 2002). Other cases of first
language loss are unintentional. Children adopted from abroad often join a
community in which there are no other speakers who share their native
language (Isurin, 2005; Nicoladis & Grabois, 2002). Even those who grow up
in a family environment where the native language continues to be spoken
may feel that the maintenance of the language is automatic, only to find
later in adulthood that, without continuous exposure, much of their prior
ability in that language have been lost (Kouritzin, 1999).
Research on first language attrition has seen a rapid increase in the past
twenty years, investigating a wide range of situations in which attrition
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
646
[645--664] 11.4.2015 6:48PM
can occur. Nevertheless, some of the fundamental questions pertaining to
first language loss continue to elude clear answers. For instance, when one
experiences native language attrition, does the effect represent a true loss
of the knowledge, of the speaker’s linguistic competence, or does the
knowledge simply become inaccessible? Studies of re-immersion in a
childhood language suggest that elements of a “forgotten” language can
become somewhat reactivated, implying that the language was not com-
pletely lost (e.g., Au et al., 2008). However, any vestiges of language with-
out re-exposure do not seem to be detected in behavioral or neuroimaging
measures (Pallier et al., 2003).
This chapter focuses on cases of native language attrition that occur
after a complete or near-complete cessation of input from that language.
While the term ‘attrition’ is sometimes used to refer only to language
change or language decline after acquisition has been completed (typically
adulthood), we would argue that the L1 is never ‘completely’ acquired, or
that there is no point in time at which one can claim it is, and thus we use
the term ‘attrition’ to refer to any decline in language abilities that were
previously demonstrated by an individual. The newly learned language
(the L2) is usually acquired to the same degree as for native speakers of that
language who had input from birth, although this is not always the case.
Several studies have explored what kind of vestiges of memory for the first
language remain in individuals who can no longer speak it, while others
have looked at whether such individuals demonstrate any sort of relearn-
ing advantage due to a re-activation of long-buried memory traces. While
the findings are not so clear with regard to what sort of influences are
important for first language loss or maintenance in these cases, the data
hint at factors such as the age at which L1 input ceased, social pressures,
and attitudinal factors, and possibly the density of one’s social network.
Finally, several theories of language attrition are considered in light of the
data presented on instances of complete or near-complete first language
(L1) loss.
29.2 Dramatic first language loss
Some of the most dramatic examples of native language loss have been
found among international adoptees. Children in this circumstance are
often brought to a new country at a very young age and enter an environ-
ment where they no longer have any exposure to their native language.
Given this abrupt shift in the linguistic environment, the child begins to
acquire the new language, often with great rapidity (Gauthier, Genesee, &
Kasparian, 2012; Isurin, 2005; Nicoladis & Grabois, 2002). When adoption
occurred at early ages, especially at less than one year old, adoptees have
been shown to perform in their new language just like their non-adopted
peers by the time they reach two to three years of age (Glennen & Masters,
646 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
647
[645--664] 11.4.2015 6:48PM
2002; Krakow & Roberts, 2003; Tan & Yang, 2005). Nevertheless, some
effects of the later age of second language acquisition of the adoptees
compared to non-adoptees can be detected using sensitive measures of
language production and comprehension. Hyltenstam et al. (2009) found
that as young adults, only one out of the four international adoptees they
studied performed like native Swedish speakers across a battery of ten
linguistic tests; the age of adoption of that individual was one year old.
Despite the fact that the others were adopted at 2 years, 4 years, and 9 years
old, they all showed some deviance from native-like patterns of perfor-
mance when tested between the ages of 28 and 33.
Subtle differences between adoptees and native-born controls have also
been reported in terms of hemispheric lateralization of the dominant
language (Rajagopal et al., 2013), While the control participants showed
clear left-hemisphere dominance for English, the adoptees demonstrated
more bilateral activation. These children had been adopted between the
ages of 9 months and 3 years old and were 6–10 years of age at testing.
While differences in hemispheric lateralization for language do not neces-
sarily imply that differences in language behavior or ultimate language
attainment exist – and for these pre-teen participants we are not given data
on proficiency relative to their monolingual peers, they do suggest that an
abrupt switch in language input at a young age may have effects on how
language is processed by the brain. This intriguing area warrants addi-
tional investigation to understand how the brain responds to these
changes during early language development. Overall, international adop-
tees are successful in acquiring their new language to a level that is
considered native-like or near-native-like with greater success associated
with earlier ages of adoption.
Does the success with which adoptees gain proficiency in their second
language necessitate native language loss? According to Pallier (2007), the
stabilization of brain networks involved in processing the native language
makes it more difficult to acquire a second language. Thus, rather than
maturational constraints on neural plasticity, it is the developmental
stabilization of the native language that interferes with second language
acquisition. If international adoptees are able to “erase” the first language
from their minds, this presumably frees up the language learning mechan-
ism to acquire the second language without first language interference. In
fact, these learners are sometimes called “sequential monolinguals” (e.g.,
Schmid, 2012) which implies that their language development is distinct
from that of early bilinguals.
One way of testing this hypothesis is by comparing second language
acquisition success in individuals who began acquiring the second lan-
guage at the same age but who differ in whether or not they experienced a
complete stop in first language use. A situation similar to this was tested by
Schmid (2012). Her corpus contained interviews from adults who had left
Germany during the Holocaust between the ages of 11 and 15 and moved
Losing an L1 to an L2 647
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
648
[645--664] 11.4.2015 6:48PM
to English-speaking countries. Some of them emigrated with their families
while others were part of a child-rescue program called Kindertransport,
which placed children in the homes of English-speaking families. The
latter group has many similarities with the international adoptees from
other studies in that the use of the native language was abruptly halted due
to both the lack of access to German speakers and the political situation at
that time, which was hostile toward German speakers. Crucially, however,
these adoptees were much older than typical international adoptees.
Schmid found no significant differences between the two groups of older
adults in their English production, measured for complexity, fluency, and
accuracy of syntax, morphology, and lexical semantics. This finding con-
tradicts the prediction made by Pallier (2007) that the abrupt stop and
subsequent forgetting of the native language facilitates second-language
acquisition. However, it must be noted that even the children who
migrated with their families stopped using German in public, with many
families switching to the use of English even at home. Thus, the length of
time that the native language continued to be used was not much greater
for the family migrants than for the Kindertransport migrants. There are
other crucial differences between the participants in Schmid (2012) and
the international adoptees, including age at migration and attitudes
toward both languages. Therefore, it is crucial that additional research
investigate the role of first language loss and maintenance in second
language acquisition before strong conclusions can be drawn.
What happens with the first learned language in the case of interna-
tional adoption? Pallier et al. (2003) tested adults who had been adopted
from Korea and grew up in France. In their twenties and thirties they were
tested on their ability to identify Korean speech from other unknown
languages such as Japanese, Polish, Swedish, and Wolof. They were also
asked to identify which of two Korean words was the translation of a given
French word. The eight Korean adoptees all considered themselves mono-
lingual French speakers and had not studied Korean or used it since they
were adopted, which ranged from 3 to 8 years of age. The participants were
no better at identifying Korean sentences among sentences in the other
unknown languages than a group of French-speaking controls, nor were
they any better than controls at identifying any of the other languages. In
fact, both Korean and Japanese sentences were rated just as likely to be
Korean by both groups of participants. The participants were then tested in
an fMRI scanner to determine whether listening to Korean sentences
produced different patterns of brain activation in the adoptees and non-
adoptees. Comparing activation patterns for French and Korean sentences
revealed no significant differences between the two groups. In fact, listen-
ing to Korean sentences produced the same activation as listening to Polish
sentences in the group of adoptees. This study reveals important findings
about the nature of language maintenance for early-learned languages
in individuals who experience an abrupt loss of input and use. Both
648 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
649
[645--664] 11.4.2015 6:48PM
behaviorally and neurologically, there appears to be no trace of the first
language that becomes reactivated when re-exposed. However, other
researchers (e.g., Hyltenstam et al., 2009) have challenged these findings,
pointing out that the method used by Pallier et al. to test residual L1
knowledge may not have been sensitive enough to detect memory traces
of the first language. They support their argument with the observations
that no differences were detected between the adoptees and native French
controls nor were any effects of age of L2 acquisition found in the group of
adoptees, factors which Hyltenstam et al. argue should have been present
in the data obtained by Pallier et al. based on other research finding robust
effects of these factors.
Another study of Korean adoptees in France tested phonological discri-
mination of Korean voiceless consonant contrasts that are difficult for
French speakers to perceive (Ventureyra, Pallier, & Yoo, 2004). Eighteen
Korean adoptees discriminated the Korean phonological contrasts no bet-
ter than French controls and far below the performance of Korean con-
trols. A subset of the Korean adoptees who had had some re-exposure to
Korean in the form of short touristic trips to Korea or short-term language
classes performed slightly better than those with no re-exposure, although
even their performance did not approximate that of the Korean controls.
These two studies suggest that the abrupt halt in the acquisition of one’s
native language at an early age (in this case, up to 9 years) results in the
complete inability to discriminate phonological contrasts in that language
or even recognize when one is hearing that language. Additionally, neural
resources recruited for processing the dominant language in adulthood
are the same regardless of whether it was the first learned language or the
second in adoptees like these.
Not all studies of international adoptees show total native language loss,
however. Montrul (2011) reported on a woman of Guatemalan descent
who was adopted at the age of 9 by a family in the USA and received no
input in Spanish for her first five years in the new country. Her level of
Spanish at the time of the study was marked by a fair degree of fluency, but
frequent errors on object marking and verbal morphology. Since the
participant had taken Spanish classes in high school and college and
made attempts to speak Spanish at her workplace, it is not clear that her
Spanish was actually retained and not relearned. The later age of adoption
of this participant may also be an important factor – though recall that one
of the participants in the Ventureyra et al. study was 9 at the time of
adoption. As mentioned previously, the Kindertransport migrants dis-
cussed in Schmid (2012) had a background similar to other international
adoptees, but they were older at migration (11–15 years old). Despite the
fact that their German input stopped abruptly when they emigrated, many
of them retained enough of the language to give the interview in German
in late adulthood (65–92 years old). In fact, seven out of the nine
Kindertransport migrants chose to speak in German for their interview
Losing an L1 to an L2 649
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
650
[645--664] 11.4.2015 6:48PM
compared to nine out of 16 interviewees who had migrated with their
families. When asked to rate the relative ease with which they could
currently speak German, three rated their German use as “effortless,”
four chose “with some effort,” and only one reported speaking German
“with great difficulty.” This, in fact, was about the same distribution of self-
perceived proficiencies found in the family migrants. However, even in
this study, the interviewees could have re-activated their native language
in adulthood.
The better retention of these migrants’ native language skills compared
to the other international adoptees could have two possible explanations.
As in the case of the Guatemalan adoptee described in Montrul (2011), the
German migrants were older at the time of adoption than the Korean
adoptees tested by Pallier et al. (2003). The German migrants were 11–15
years of age at the time of emigration while the Koreans were 3–8 years old
when adopted. The conflicting results seen between these two studies are
consistent with predictions made by the Critical Period Hypothesis, which
proposes that a decrease in the brain’s neural plasticity around the time of
puberty results in decreased ability to acquire language skills with native-
like proficiency. The stabilization of neural networks at this age would also
presumably lead to less language attrition if input ceased after this stabi-
lization point. Another possible explanation, however, could be the role of
continued L1 input or re-exposure to the language. While Pallier et al.
(2003) specifically excluded Korean participants who had had exposure
to Korean since their adoption, Schmid (2012) does not provide informa-
tion regarding the extent to which interviewees were exposed to German
since their emigration or used it in their adult lives, and Montrul’s (2011)
participant reported having had re-exposure through high school and
college Spanish classes. Thus, it is just as plausible that re-exposure to
the native language in adulthood helped these participants recover certain
language elements that had attrited after input ceased.
29.3 Relearning advantages for a “lost”language
If international adoptees have been reported to show an advantage in re-
learning their native language compared to learners with no previous
exposure to the language, then we must conclude that the language had
not been altogether lost. This long-dormant residual knowledge may not
be easy to tap into, however. Recall and recognition tests may not be able
to access buried fragments of memory. Nelson (1978) argued that a ‘sav-
ings’ paradigm should be able to reveal the presence of less accessible
stored memories by identifying relearning advantages of previously
learned items over new items. In fact, de Bot and Stoessel (2000) employed
this paradigm to investigate L2 retention and found evidence for residual
vocabulary knowledge in two individuals who had not had exposure to a
650 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
651
[645--664] 11.4.2015 6:48PM
language learned in childhood for more than thirty years. Other studies of
savings in relearning L2 vocabulary after input has ceased have shown
similar findings in terms of a relearning advantage for previously known
words, though this effect appears to depend on the length of time without
input (up to forty-five years in some cases), the age at which input in that
language ceased, and the proficiency that was previously attained in the
language (de Bot, Martens, & Stoessel, 2004; Hansen, Umeda, & McKinney,
2002; Ioup, 2001; Tomiyama, 2001; van der Hoeven & de Bot, 2012). While
these studies looked at savings in L2 knowledge rather than the native
language, the application of the savings method for the retrieval of long-
unused language knowledge may prove useful for probing L1 residual
memory.
When the re-exposure to one’s native language occurs in childhood, the
findings point to a rapid reacquisition of the language. Three case studies
of young children who spent up to fifteen months abroad demonstrate the
rapidity with which children at this age become dominant in the L2 and
then again in the L1 when re-exposed to it. The first case (Berman, 1979)
chronicles the transition of a native Hebrew-speaking child with some
passive comprehension in English starting at the age of 2; 11. During the
year she spent living in the USA with her family, she spoke only the L2,
English. Upon returning to Israel at age 3; 11, the child began recovering
her use of the native language within a month and was considered an L1-
dominant bilingual by about four months after returning to the home
country. The second case illustrates a more dramatic example of a switch
in language dominance. Slobin et al. (1993) describe a native English-
speaking child who, at the age of 2; 6, moved to Turkey with his family
for a stay of five months. While the child maintained comprehension of
English during his stay in Turkey, he spoke only Turkish. The reacquisition
of his English skills upon returning to the USA was dramatic; within two
months he could no longer communicate in Turkish and was considered to
be, once again, an English monolingual. Finally, Dahl et al. (2010) docu-
ment the case of a child who was a balanced Norwegian–English bilingual
at the age of 2; 11 when he moved to the USA for fifteen months, during
which time he rarely heard Norwegian. Upon returning to Norway, the
child communicated for the first two months mainly by code-switching
and code-mixing English and Norwegian, speaking more Norwegian with
his peers at preschool and using more code-mixing patterns with the
adults at the preschool, who understood both languages, but in either
case rarely producing utterances only in Norwegian. At week 8, there
was a sudden shift in the child’s communication patterns to mostly
Norwegian utterances and by week 11, the child no longer code-switched.
It is possible that the reacquisition process is less rapid in older children
who are re-immersed in their native language after a time away from it.
One report of an adolescent who spoke three languages quite fluently until
the age of 6 and then spent six years abroad between the ages of 6 and 12
Losing an L1 to an L2 651
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
652
[645--664] 11.4.2015 6:48PM
during which time he only communicated in his newly learned language,
English (Faingold, 2004). At age 12, he was re-immersed in two of his native
languages (Hebrew and Spanish) through classes at school and is reported
to have become proficient in them again quite quickly, which the author
attributes to positive attitudes toward the languages, which are tied to the
student’s cultural heritage. Unfortunately, no quantitative data is available
for this participant to ascertain to what extent and at what rate he was able
to recover proficiency in the two native languages. However, another
study involving three adolescent siblings does provide data assessing lan-
guage skills before emigration, after a period of two years nine months in
the new country, and again seven months after returning to the native
language environment (Hubbell-Weinhold, 2005). The three children were
aged 11; 5, 10; 2, and 8; 7 at the time of their emigration from the United
States to Switzerland. When tested almost three years after immigrating,
all three children showed large declines in both receptive and expressive
language. Four months later, they returned to the USA at which point
German input almost entirely ceased. When tested in English seven
months later, their English skills in all areas tested had recovered but
only by 20–45 percent of their original language scores. These case studies
demonstrate that reacquisition of a native language is typically quite rapid
when it occurs in early childhood, though from the limited set of data we
are able to examine, it appears that either the age of immigration and
reimmersion or the duration of the time without L1 input may affect the
speed with which one is able to regain attrited language skills, though they
may both play important roles.
Residual memory of a native language from childhood is less apparent
when individuals are tested as adults, though many of the studies of adults
investigated either the presence of retained language abilities without
relearning or abilities manifested in very early stages of relearning, mostly
in terms of phonetic discrimination. As mentioned previously, Ventureyra
et al. (2004) found that Korean adoptees in France performed far below
50 percent accuracy in discriminating among Korean consonants that
were not distinguishable to their French-speaking peers, and, indeed,
their performance did not differ from French monolingual controls.
However, a very similar study of Korean adoptees to the USA showed a
clear advantage by Korean adoptees on two of the three consonant types
(plain and aspirated, but not tense) over novice Korean learners after only
two weeks of college-level beginner Korean classes (Oh, Au, & Jun, 2010).
Unlike the previous studies of adoptees, all but one of the adoptees in the
Oh et al. study were adopted by the age of 1.
Oh et al.’s (2010) findings are supported by other studies showing that
early language exposure without continued use may confer some long-
lasting benefits in speech sound discrimination. Tees and Werker (1984)
found that English monolingual adults who had been exposed to Hindi in
early childhood (until 1–2 years of age) showed significantly better
652 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
653
[645--664] 11.4.2015 6:48PM
performance distinguishing a Hindi dental/retroflex contrast compared to
adults without early Hindi experience when they were tested as young
adults only seven to ten days after the start of a Hindi college class for
beginners. In fact, their scores were similar to those of a group of English–
Hindi bilinguals with more than five years’ experience studying Hindi and
still much better than the scores obtained a year later by the Hindi learners
without early exposure.
In another college-level first-year Korean language class (Oh et al., 2003),
participants were tested who had been exposed to Korean in early child-
hood, but experienced a rapid decline in Korean input before the age of 7.
Based on their language background questionnaires, they were divided
into those who overheard Korean as youngsters but rarely spoke it and
those who spoke Korean regularly for at least three years. On the phoneme
perception task, which included the same trio of stop consonant contrasts
as the previous Korean studies, childhood speakers and hearers performed
similarly to native speakers, outperforming novice learners. By contrast,
only the childhood Korean speakers and native speakers reliably distin-
guished between all three consonants in their speech production, while
childhood Korean hearers and novice learners did not differentiate them
in speech. Overall accent ratings for childhood speakers were higher than
for childhood hearers and novice learners, though they were still lower
than native speakers. This benefit for accent is also reported for English-
dominant childhood Spanish speakers in a Spanish language learning
setting as adults, in which both childhood Spanish speakers and hearers
spoke Spanish with a better accent than novice learners (Au et al., 2008).
The childhood speakers showed better performance than hearers and
novice learners on grammatical measures. By contrast, Au et al. (2002)
found that adult Spanish childhood hearers performed better on measures
of Spanish phonetic production and overall accent, but no better than
novice learners on measures of morphosyntax.
These studies suggest that a slight relearning advantage may exist for
adults who had childhood exposure to the language, but most tested
these adults at the very early stages of re-immersion. Does continued
learning show even greater facilitation effects for these learners?
Bowers, Mattys, and Gage (2009) suggest that this might be the case.
They asked English-dominant speakers who had been exposed to Hindi
or Zulu as children to perform a same-different task on a set of Hindi and
Zulu phonemic contrasts that are difficult for English speakers to per-
ceive. Compared to a group of controls, the “relearners” showed signifi-
cantly better performance on these contrasts, but only for the
participants who were under age 40 at time of testing. In addition,
evidence of improved learning only became apparent after fifteen to
twenty sessions of training. Thus, studies testing participants with no
re-exposure or at very early stages of relearning may fail to find any
advantage (e.g., Pallier et al., 2003).
Losing an L1 to an L2 653
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
654
[645--664] 11.4.2015 6:48PM
One study importantly suggests there are individual differences among
individuals who have lost their L1. Hyltenstam et al. (2009) compared the
performance of two groups of college-level students studying Korean: one
group consisted of Korean adoptees to Sweden and the other included
native Swedish speakers. The participants were tested on a phonological
discrimination task of Korean phonemes differing in VOT length and a
grammaticality judgment task testing case morphology, verb morphology,
and adverb placement. The native Swedish speakers scored significantly
higher than the adoptees on the grammaticality judgment task.
Considering the fact that the Swedish natives had spent more years study-
ing Korean than the adoptees and more time living in Korea, this result was
not particularly surprising. Despite the differences between groups in
length of time learning, both groups scored similarly on the phonological
discrimination task. The authors note, however, that the range of scores by
the Korean adoptees was quite a bit greater than that seen for the native
Swedes and that about one-third of the Korean adoptee participants
demonstrated scores on this task that were higher than any of the native
Swedish speakers. In fact, the two highest performers were the two parti-
cipants with the oldest age of adoption (10 and 9 years), lending support to
the idea that re-learning of this phonological distinction could have been
facilitated in the group of Korean adoptees, particularly for those with a
greater number of years of exposure to Korean before adoption in Sweden.
Nevertheless, the advantage is slight and could only be seen on the pho-
nological perception task. In addition to the high-scoring adoptees,
another one-third of the Korean adoptees performed below any of the
native Swedes on this task. These were not necessarily those who were
youngest at age of adoption, however, so it remains a curious finding.
Moreover, without appropriate pre-tests before re-immersion in the native
language, we cannot adequately distinguish what has been retained from
what has been relearned, as is the case for several of the other studies.
Most of the studies investigating re-immersion in a native language
involve participants who were children when they immigrated, but a
single case study of an adult with a long period of no L1 exposure investi-
gated recovery of first language skills through re-exposure during a series
of interviews over four years. Stolberg and Mu¨ nch (2010) report on a
German immigrant who had moved to the USA at the age of 28 and had
been living there for almost fifty years. The researchers gathered conversa-
tional data from the informant fifteen times over the course of four years.
Throughout this time, there was a noticeable increase in fluency as well as
a decrease in the number of lexical–semantic, syntactic, and morphologi-
cal errors. The lexical–semantic errors made up a greater proportion of
errors at the beginning and also showed the greatest improvement over
time. Syntactic errors appeared to fluctuate quite a bit from session to
session with a slight overall tendency to decrease. Morphological errors, by
contrast, were quite stable overall, showing a barely decreasing tendency.
654 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
655
[645--664] 11.4.2015 6:48PM
Since this participant was not taking German language classes in order to
brush up her language skills, but rather was engaged in increased commu-
nicative exposure through the interview sessions, this seems to provide
clear evidence of a certain “reawakening” of long-forgotten native lan-
guage skills. In the end, there was not what the authors would consider a
complete recovery of native language functions. Thus, it remains to be
seen whether those aspects are forever lost or if continued exposure would
also bring those back.
29.4 Recovering a lost language through hypnosis
The contradictory findings in the literature on relearning advantages
might have to do with the presence or absence of re-exposure to the native
language, as we have suggested above. In the case of Pallier et al. (2003) and
Ventureyra et al. (2004), most of the participants were not enrolled in
second language classes, while in the rest of the cases they were, even if
only for two weeks. The one participant reported in Ventureyra et al.
(2004) who had taken Korean language classes for a few months prior to
the testing showed better performance on the phonological contrast than
the rest of the adoptee group. Could it be that a long-forgotten language
from childhood simply needs to be “turned on” again with a little stimula-
tion? If this were the case, then other forms of “reactivation” should be
able to stimulate the memories of a long-unused language as well. One
technique that has been tried is hypnosis. Only a few researchers have
published case studies on this technique. As (1962) reported on a native
Swedish speaker who moved to the USA with his family at the age of 5 and
began to speak English exclusively by the age of 7. Under age-regression
hypnosis, the participant was able to answer “more” questions in Swedish
correctly. In a more dramatic case, a Japanese-American man who claimed
to not be able to speak Japanese was suddenly able to speak it rapidly
during the psychiatric session with the author when under age-regression
hypnosis that brought him back to 3 years of age (Fromm, 1970).
A third study comes not from the recovery of a forgotten native lan-
guage, but rather a second language learned early in childhood. Footnick
(2007) reports on a native French speaker who spent 3.5 years in Togo at an
early age (2.5–6 years old) and spoke fluent Mina (a dialect of the region).
Upon returning to France, the participant was told not to use Mina any-
more so as not to hinder his development in French, and thus the partici-
pant had not used the language in fifteen years and could only produce
about thirty isolated words at the time of the study, but could not partici-
pate in conversations. While the first four hypnosis sessions resulted in
only limited responses by the participant in Mina, in the fifth session he
responded in Mina 72 percent of the time and spoke in complete, gram-
matical sentences. Upon listening to the recordings after the hypnosis
Losing an L1 to an L2 655
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
656
[645--664] 11.4.2015 6:48PM
session, the participant was astonished to hear himself speaking fluently
in Mina, and when asked to translate what he said to French, was unable to
do so. Interestingly, by the sixth session, the participant was able to
comprehend many more questions in Mina out of hypnosis than pre-
viously, though still less than what was observed during hypnosis. The
author suggests that there is a distinction between a “hidden language”
which is not lost but rather, inaccessible, and a “forgotten language”
which is no longer able to be recovered by any means. She proposes that
in a situation in which a language is ‘actively’ forgotten, i.e., actively
suppressed, hypnosis may help by removing the conflict associated with
that memory and allowing the mind to access it. However, in a case where
the language is truly “forgotten,” hypnosis would have no effect and the
language would not be recoverable.
The studies on hypnosis bring us back to the central question posed in
the introduction: can a language learned in early life be completely lost or
does it just become inaccessible? Bjork and Bjork (2006) offer a distinction
between storage strength and retrieval strength. They posit that informa-
tion that becomes entrenched in long-term memory storage remains there
and does not decay, and that it is the retrieval strength that decays over
time due to non-use. This theory would predict that storage strength would
increase as opportunities to practice increase, which should be directly
related to amount of input and number of years of exposure before that
exposure is broken. Thus, age of acquisition would predict the level of
entrenchment of the first language. Second, the theory would predict that
re-immersion would increase the language’s retrieval strength, though it is
not clear whether the process is item specific or generalizable.
29.5 Factors in first language loss and maintenance
The rate at which native language proficiency declines has often been
observed to be quite rapid in children. Several longitudinal studies of
immigrant children give us a glimpse into the rapid transition from
being dominant in one language to becoming highly fluent and dominant
in a new language. Nicoladis and Grabois (2002) followed the language
development of a Chinese child adopted into an English-speaking
Canadian home at the age of 17 months. Between the ages of 1;6 and 1;9,
the child was observed ten times, five times during English-speaking
playtime and five times during Cantonese-speaking playtime with a native
speaking play partner of each language. At the first session, conducted four
weeks after the child’s arrival, the girl produced three utterances in
Chinese and only one in English. In subsequent sessions, the child only
demonstrated comprehension in Chinese but did not attempt to produce
it, and even evidence of comprehension ceased by the fourth session. In
contrast, starting with the second session, the child began producing a
656 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
657
[645--664] 11.4.2015 6:48PM
number of utterances in English and also showed an increase in compre-
hension abilities and produced imitations. Thus, less than six weeks after
arrival, the child had stopped producing utterances in the native language
and by three months, she no longer showed evidence of comprehension.
Kaufman and Aronoff (1991) describe a similarly rapid switch in lan-
guage dominance, though a period of bilingual code-mixing seems to
lengthen the process somewhat. In this study, a Hebrew-speaking child
who immigrated with her family to an English-speaking environment at
the age of 2;6, went through several stages in the process of switching
language dominance. Unlike the case study just described, this child
remained within a Hebrew-speaking family environment but began
attending an English-speaking nursery school soon after arriving. The
first stage of Hebrew attrition began just three months after immigrating
and was characterized by lexical borrowings from English which were
inflected with Hebrew morphosyntactic elements. This stage lasted for
about four months, after which there was a very brief stage of balanced
bilingualism (only two months), in which the child exhibited a high level
of fluency in both languages in the appropriate environments. Starting at
age 3;2, the Hebrew production of the child began to show signs of notice-
able decline at home, although her parents continued to speak only
Hebrew with her. This was coupled with a change in attitude toward her
native language and a refusal to speak Hebrew at times. In the final stage of
language change observed in this study (from 3;5–4;6), about a year after
her immersion in English, the child used a simplified version of Hebrew
morphosyntax in which all verbal paradigms had been reduced to a single
idiosyncratic verbal template. Decline in L1 use, then, may be more or less
abrupt depending on the circumstances and environment and, likely, on
individual differences as well.
In particular, we might consider the young age of some of the partici-
pants described in this section to accelerate their shift in language dom-
inance. Given the reduced input of the native language at such an early
stage in language development (which was more severe in the case of the
Cantonese-speaking child than the Hebrew-speaking child), there might
be fewer opportunities for these children to gather evidence, both positive
and negative, that would allow them to continue to develop native lan-
guage capacities. However, this suggestion predicts only a stagnation of
development – as Merino (1983) reported for first language comprehen-
sion – not a reversal. As was seen in several of the case studies described
earlier (e.g., Berman, 1979; Hubbell-Weinhold, 2005; Slobin et al., 1993), a
reduction of L1 input results in a decline in L1 language skills rather than
simply a lack of development. Indeed, this was even seen in older children
(Isurin, 2000, 2005; Ventureyra et al., 2004), whose L1 can be said to have
been quite developed by the time their exposure to the language stopped.
Another important factor in language maintenance may be attitudinal
changes. A number of studies involving children note the participants’
Losing an L1 to an L2 657
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
658
[645--664] 11.4.2015 6:48PM
reluctance or unwillingness to continue speaking the native language,
even when the opportunity to interact in this language still exists (e.g.,
Isurin, 2005; Kaufman & Aronoff, 1991, Leopold, 1939–1949). This may be
due to social pressures or the desire to fit in with a new social environ-
ment. Schmid (2012), for example, mentions that her interviewees noticed
some loss of their first language “within a few months” of arriving in their
new country. Her participants were German speakers immigrating to
England or North America just before the start of the Second World War,
so their motivation for not using or appearing German was strong. Thus, it
seems that the attentional resources toward second language acquisition
and the implicit social pressures to learn a new language may be key
contributing factors in language loss. Indeed, quite rapid language loss
has also been observed in older children. For instance, Isurin (2000, 2005)
observed changes in lexical retrieval efficiency and word order patterns
within the first year after adoption by a Russian girl who was adopted at
the age of 9 by an American family. When tested two years after her
adoption, she was able to successfully retrieve less than 40 percent of the
Russian words she was tested on. The participants in Schmid (2012) were
between the ages of 11 and 15, though the rate of decline in their native
language skills is only anecdotal.
On the other hand, there do seem to be important effects of age of
immigration on native language retention. As mentioned previously, the
case study described by Stolberg and Mu¨ nch (2010), a woman who immi-
grated at the age of 28, showed that even after fifty years of nearly com-
plete disuse, the participant’s speech was relatively fluent, though prone
to self-interruptions and a variety of deviations from the norm.
Nevertheless, a similar situation in younger children appears to have a
much greater impact on native language retention. The relationship
between age of immigration and native language maintenance is not
exactly straightforward, however. While some studies show a small degree
of native language maintenance or relearning advantage in individuals
whose input ceased by just one or two years of age (Oh et al., 2010; Tees &
Werker, 1984), others show no retention even for those who spoke the
native language exclusively until the age of 8 or 9 (Pallier et al., 2003;
Ventureyra et al., 2004). Ko¨ pke and Schmid (2004) suggest that stabiliza-
tion of the native language linguistic system – which occurs around pub-
erty – is necessary to prevent dramatic language loss. However, this
proposal does not account for the retention of certain native language
features in individuals with early cessation of language exposure.
Additionally, some individuals who emigrated before puberty are able to
retain enough language skills to be interviewed in the language as young
adults (Flores, 2012), though what may be termed errors (since they were
surely mastered prior to emigration) are relatively frequent.
Continued exposure to a language seems to have a positive effect on
language retention, as can be seen in the general better language outcome
658 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
659
[645--664] 11.4.2015 6:48PM
of heritage language speakers compared to international adoptees (e.g.,
Kenji & D’Andrea, 1992; Mikulski, 2010; Oh et al., 2003; Tran, 2010).
Nevertheless, it is interesting to note that rapid declines in native language
processing were found both in cases of adoption, where children often had
no one to converse with in their native language, as well as in cases of
familial migration, where children still had opportunities to speak their
native language, if only with other family members (e.g., Hubbell-
Weinhold, 2005; Kaufman & Aronoff, 1991; Slobin et al., 1993).
For example, Schmid (2012) found almost no significant differences
between Kindertransport migrants and family migrants when they pro-
duced German later in life, nor were there significantly fewer participants
in the Kindertransport group who chose to speak German for the inter-
view. Oh et al. (2010) and Ventureyra et al. (2004) also found no significant
differences between those who had had exposure after immigration and
those who had not. What may be most important for preventing first
language loss is a certain minimal level of continued exposure. The ques-
tion of just how much is yet to be explored.
As well, the density of one’s social network has been reported to be an
important factor in the degree of adult first language attrition (Hulsenet al.,
2000; Schmid, 2012) as well as in the amount of first language development
in heritage language speakers (Tran, 2010), and it may also be an important
factor in first language loss or maintenance in children who immigrate to a
place where there are no first language connections outside the home.
Additionally, age of second language acquisition and amount of language
input are often confounded, making it difficult to determine what role each
of these factors plays in both second language acquisition and first language
attrition. For instance, Jia and Aaronson (1999) provide evidence that
younger immigrants were more likely to become dominant in the second
language than adolescent immigrants. However, younger participants also
received more years of second language input (and, presumably, less native
language input) than older immigrants. These issues need to be separately
examined in order to understand their contributions to first language loss.
Some researchers argue that certain aspects of language are more vul-
nerable than others to language attrition. Since few of the studies
described in this paper looked at language loss longitudinally, it is not
clear whether the loss of L1 skills occurs at different rates for different
aspects of language. Most studies that were longitudinal in nature con-
strained their analysis to only specific areas of language (morphosyntax in
Kaufman & Aronoff, 1991, lexical retrieval in Isurin, 2000, and word order
in Isurin, 2005) or only described general language use preferences
(Berman, 1979; Faingold, 2004; Slobin et al., 1993). Hubbell-Weinhold
(2005), however, included tests of both semantics and syntax, as well as
including both receptive and expressive language measures, and found
that semantics showed a greater decline than syntax, while expressive and
receptive language skills showed a similar degree of decline. On the other
Losing an L1 to an L2 659
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
660
[645--664] 11.4.2015 6:48PM
hand, Isurin (2000) notes that her participant became reluctant to partici-
pate in storytelling and picture description tasks at later testing sessions
but continued to perform the lexical retrieval tasks, implying that she felt
more comfortable with lexical retrieval than with syntax. Within syntax,
Flores (2012) found that overt pronoun realization was more vulnerable to
attrition than word order constraints. An interesting observation has been
made noting that similar errors are seen in the production of heritage
language speakers and in second-language acquirers and that these errors
may indicate certain aspects of grammar that are both harder to acquire in
second language acquisition and easier to lose in first language attrition
(Montrul, 2005). Such parallels are intriguing and warrant further study.
A hypothesis presented by Sorace (2005) proposes that aspects of “pure”
syntax are not vulnerable to attrition once acquired since they are proce-
duralized, but that syntactic constructions that incorporate discourse,
lexical, or pragmatic information (so-called “interface” phenomena) are
more likely to undergo attrition. The consolidation of purely syntactic
components is thought to occur around the age of puberty, predicting
that the cessation of L1 input before and after puberty will result in
qualitatively different patterns of language loss, a prediction that has
received some support (Flores, 2010, 2012). It is not so clear how this
applies to early cases of first language attrition, and the question of
which facets of language tend to be lost first in situations of total cessation
of input is yet to be explored.
Age of immigration, then, social reasons to lose the L1, and possibly the
number of social contacts in the L1 appear to enter into whether the L1
attrites, and into the rate at which it does. Syntactic complexity and
whether or not certain structures extend beyond ‘pure’ syntax may deter-
mine which structures are ‘lost’ first. Other factors, like interference into
L1 from L2 (see Higby, Kim, & Obler, 2013) have not been explored in much
detail in the literature on dramatic first language loss, though Isurin (2000)
notes that the retention of L1 words shows a strong relationship with the
acquisition of their L2 translation equivalents in that the L2 words that
were learned last are the translation equivalents of the L1 words that were
retained the longest. Given that L2 interference has been found to play an
important role in the general research on L1 attrition (e.g., Cook, 2003),
this relationship should be explored in more depth in cases of early first
language loss to discover in what ways language attrition in younger and
older individuals may differ.
29.6 Theoretical and neurological explanations
of language loss
A number of theories have been proposed to account for the patterns seen
in research on first language attrition. For the most part, these models
660 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
661
[645--664] 11.4.2015 6:48PM
were developed to account for changes in L1 language patterns or proces-
sing in bilinguals rather than to explain such dramatic L1 loss as what has
been reviewed in this chapter. Nevertheless, we consider how they might
handle the set of findings we have described.
One set of these theories assigns the process of second language acquisi-
tion as the driving force of attrition seen in the first language. For instance,
the Speech Learning Model (Flege, 1995, 2002) proposes that the first and
second languages exist in the same “phonological space” and therefore
mutually influence each other, and that language-specific speech categor-
ization mechanisms remain intact throughout the lifespan after the first
year. Language input and language dominance, then, can influence speech
perception and production in the native language by adapting these to
second language categories. There is limited evidence for this theory in the
data described in this chapter, however. While some individuals showed a
‘savings’ advantage when relearning certain phonological contrasts in
their native language (Oh et al., 2003, 2010; Tees & Werker, 1984), others,
some of whom had had exposure to the native language for much longer
than one year did not (Bowers et al., 2009; Hyltenstam et al., 2009; Pallier
et al., 2003).
A second proposal, mentioned earlier in the chapter, is the Language
Replacement Hypothesis (Pallier et al., 2003). This hypothesis states that
the first language serves as a barrier to successful acquisition of the second
language, and therefore the extinction of the native language allows the
second language to be fully acquired and to replace the first as the person’s
‘native’ language. Related to this is Footnick’s (2007) proposal that the
process of ‘active forgetting’ of the native language is key in first language
loss and may occur due to environmental pressures or the suppression
of negative associated memories. Recently, the role of active inhibition has
been implicated in short-term losses in first language abilities and
associated second language gains (Linck, Kroll, & Sunderman, 2009).
After native English speakers spent three months studying in a Spanish-
speaking country, they produced significantly more responses on a
Spanish semantic verbal fluency task than a comparable group of
Spanish learners who stayed in their home country. The immersion in
the second language seemed to have made the first language less accessi-
ble, however, as the immersion students produced significantly fewer
responses in the native language than their counterparts did.
Given that many adult balanced bilinguals exist, Pallier’s proposal thus
appears to be too strong to account for data showing successful second
language acquisition in the face of first language maintenance. However,
the idea that active suppression of the native language may be involved in
the cases described in this chapter is not implausible. If children find that
their native language is no longer useful and experience social pressures to
acquire the new language quickly, inhibition of the native language may
be a useful strategy. Evidence for this comes from Isurin’s (2000) finding
Losing an L1 to an L2 661
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
662
[645--664] 11.4.2015 6:48PM
that response times for native language lexical retrieval increased across
the testing sessions (which may indicate the inhibition of these items) and
that high frequency L1 items and cognate words were more likely to
evidence difficulty for retrieval than low frequency, non-cognate words.
Another set of proposals places a stronger emphasis on L1 language use
for language maintenance. The Weaker Links Hypothesis (Gollan et al.,
2005; 2008) proposes that reduced use of a given language leads to weaker
links between the semantics and phonology of its lexical system. A similar
account is the Activation Threshold Hypothesis (Paradis, 2007). ‘Activation
threshold’ refers to the resting-state activation level of language items that
determines how much effort is needed in order to activate them for
language use. Items with a higher threshold require more processing
resources in order to activate them, and activation thresholds change
over time according to language-use patterns. The hypothesis predicts
that native language disuse will lead to higher activation thresholds and
that highly frequent items in the second language will replace their coun-
terparts in the native language with time. The Dynamic Model of
Multilingualism (de Bot, Lowie, & Verspoor, 2007; Herdina & Jessner,
2002) suggests that language is a dynamic system, in which limited
resources must be applied to the maintenance of all languages. Thus,
increased use of the second language leads to natural decay of the first
language because of the increased effort required to maintain it. Each of
these theories is potentially applicable to the studies of complete and near-
complete first language loss if extended to their extremes. For instance, if
the links between semantics and phonology weaken, do they eventually
weaken to the point where they are “broken” making access to the pho-
nological shapes of words impossible? Studies of savings in relearning
previously known words imply that weakened connections can be
restrengthened with training. This presumably applies to the first lan-
guage as well. The Dynamic Model of Multilingualism may be able to
account for the particular vulnerability of younger children to rapid first-
language decline if we assume that the resources needed to maintain
languages in an active state are even more limited for younger children
than for adolescents and adults.
Another aspect of a dynamic systems perspective on language attrition
involves the primary role of social adaptation. In their description of
language as a complex adaptive system, Beckner et al. (2009) place the
central force driving language acquisition and language change on social
interactions. In this view, adaptation of the linguistic system to environ-
mental factors such as input and social interactions continues throughout
one’s lifetime. Language input factors, such as form frequency and the
continuous co-occurrence of form, meaning, and usage patterns of a parti-
cular construction, are essential for both language acquisition and lan-
guage maintenance (de Bot et al., 2007; Ellis & Larsen-Freeman, 2009).
Indeed, social interactions and attitudes toward languages appear to play
662 EVE HIGBY AND LORAIN E OBLER
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
663
[645--664] 11.4.2015 6:48PM
an important role in language maintenance especially for children and
adolescents (Kaufman & Aronoff, 1991; Kouritzin, 1999; Slobin et al.,
1993).
How these theories might incorporate those aspects of neurology that
must be called upon to explain L1 loss and attrition has not yet been
postulated beyond speculation. Certain of the terms in the theories
mentioned above imply neurological notions, though they could, in prin-
ciple, be strictly psychological constructs: the Language Replacement
Hypothesis of Pallier and colleagues is consistent with cortical and sub-
cortical regions structured for language learning and maintenance being
called upon for L2 learning in ways that displace networks and/or reduce
synaptic strengths associated with the L1. Footnick’s notion of active
forgetting and that of Linck et al. of active inhibition might bring about
similar neurological consequences whereby networks and pathways used
in L1 are taken over by L2. Paradis’ (2007) activation thresholds must also
involve neural substrates. The notion of L1 having achieved a certain
degree of stability in order to resist attrition must reflect underlying neural
pathways, perhaps including myelination and chemical processes as well.
Plasticity would interact with any and all of these neurological changes, in
dynamic ways (as the model of de Bot and colleagues suggests) as one
language dominates and then another one does, and daily frequency of
input and output appear to enter as well.
29.7 Conclusion
What is perhaps most surprising about first language attrition is the range
of findings that have been reported. In some adults, there is no evidence of
an L1 that was learned up until adoption as late as age 9; in other children
who immigrate to an L2 as young as 1–2 years of age, the ability to still
discriminate some of their native language sound contrasts remain acces-
sible even if only implicitly. It is not surprising that age of immigration
enters into determining the relative end-state achieved for L1 in such
situations, since relatively less input of L1 to a young brain will under-
standably result in less well-consolidated mastery of it. Moreover, aspects
of the language that had not been learned at the time when a sharp
increase in L2 exposure occurs may never be acquired, or, if they were
only partially learned but not fully mastered may attrite in the child. What
remains unclear are what role attitude toward the L1 (and L2) plays in
determining the amount of attrition or loss, and the extent to which a ‘lost’
language might be stored, hidden but inaccessible. Virtually unstudied at
this time are individual differences in language-learning abilities, or cog-
nitive abilities that underlie them, that may determine how two indivi-
duals in very similar situations of L1 and L2 exposure may show more or
less L1 attrition, as we know two individuals in very similar L2-learning or
Losing an L1 to an L2 663
C:/ITOOLS/WMS/CUP-NEW/6140802/WORKINGFOLDER/SWIET/9781107060586C29.3D
664
[645--664] 11.4.2015 6:48PM
acquiring situations may show different degrees of success in L2 acquisi-
tion (e.g., Novoa, Fein, & Obler, 1988). From the number of serious theories
proposed to account for subsets of the language attrition data, it appears
this field is well into its early adolescence. With more neurological studies
to follow, greater precision of our understanding of the biological and
psychological aspects of first language loss should evolve.
664 EVE HIGBY AND LORAIN E OBLER