6

Child second language,

multilingual and heritage
language acquisition,
language attrition

6.1 Adult L2 acquisition 142 6.6 Commonalities and differences

6.2 Child L2 acquisition 142 between the four acquisition
6.3 L3/Ln acquisition 147 contexts 161
6.4 Heritage language learners 153 6.7 Exercises 168
6.5 Language attrition 158

Starting in the previous chapter and in this one, we are thinking about
individuals with two or more languages in their brain. There is one issue
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

that unites all bilinguals and multilinguals, no matter when the other
languages were learned: their minds/brains are not identical in their func-
tionality to those of monolinguals, precisely because the former operate
with two and more languages. This fact is of paramount importance when
discussing multilingual linguistic behavior. It may confer a broader cogni-
tive benefit, but in day-to-day language use, it may also make lexical access
harder and processing more taxing.
In this chapter, I will look at various language acquisition and change
conditions. I will briefly introduce the process of adult L2 acquisition,
which is the focus of this book. Next, I compare child and adult L2, to see

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 142 6 CHILD SECOND LANGUAGE

if these are qualitatively different processes. I will then present L3 or Ln

language, or multilingual, acquisition. Heritage language speakers are
bilingual speakers who are using their native language while speaking
another, dominant language. Language attrition, the loss of the native
language, may arise under the conditions of bilingualism even if speakers
are late bilinguals. All of these conditions of incrementally adding or
losing a language will help us think about the important factors affecting
these changes. This chapter paints a multifactorial picture of language
acquisition. At the end of this chapter, you will be able to appreciate some
of the important factors and variables that researchers take into account in
order to investigate the process. But first, let me introduce adult native
speakers.

6.1 Adult L2 acquisition

Who can be considered an adult L2 acquirer? The current cutoff point for
considering someone an adult L2 learner is 7 or 8 years of age, based on
Johnson and Newport’s classical (1989, 1991) studies. However, there has
been some variation in this respect in the literature. For example, Lenne-
berg proposed 6 years of age to be the end of the Critical Period for
language acquisition. Another commonly used cutoff point is puberty
(between 11 and 14 years of age nowadays). As the rest of this textbook
deals with the way adults acquire a second language, we will not spend
too much time in this section describing acquisition routes. However,
remember that the conditions of acquisition (in the country where the
language is spoken or not), type of exposure (naturalistic or classroom),
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

and length of language exposure and use may matter at least as much
as age for the convergence of the speakers on a nativelike mental
grammar.

6.2 Child L2 acquisition

When the initial exposure to the nonnative language is approximately

between the ages of 4 and 7, we define it as child L2 (cL2) acquisition
(Schwartz 2004, Meisel 2011). The reasoning for the lower limit is that 4 is
considered the age when the bulk of the native grammar is roughly in

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.2 CHILD L2 ACQUISITION 143

place.1 The age of 7 or 8 has been argued to be the critical age after which
acquisition may not proceed as in child L1. Although opinions in the
literature vary, there is some experimental support for these cut off ages.
For example, the classic studies by Johnson and Newport (1989, 1991)
identified children who started to acquire the L2 before this age as most
likely to fall in native ranges of competence.
My approach here will follow Bonnie Schwartz’s groundbreaking work
on this topic (Schwartz 1992, 2004, 2009, see also Lakshmanan 1995). She
calls the child–adult L2 comparison “the perfect natural experiment”
(Schwartz 2004: 17). Why would that be? What is the rationale for consid-
ering L2 children such an important population, with respect to the UG
issue? Recall from Chapter 4 that, according to the Critical Period Hypoth-
esis, second language acquisition cannot proceed as child language acquisi-
tion, if the learner is already an adult (aL2). This logic can be extended as
follows: if the learner is not an adult, that is, between the ages of 4 and 7 or
8, L2 acquisition should proceed differently. For example, in Johnson and
Newport’s (1989) study, the child acquirers were generally more successful
than those who started as adults, and a negative age-to-nativeness correl-
ation was observed. Thus, if research uncovers evidence that the develop-
mental paths of L2 children and adults look the same, in the sense that they
make similar errors and exhibit the same developmental stages, this would
constitute evidence that they are both going the UG-sanctioned route. If, on
the other hand, the developmental course of the adult L2 learner diverges
from that of the child L2 learner, this would be evidence that adult L2
acquisition is not UG-constrained and may be due to rote learning, various
problem-solving strategies, and superficial noticing of linguistic rules. The
L1–cL2–aL2 comparison, then, can address a seminal research question,
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

namely, the question that we have been pondering in this textbook from the
very beginning.
What are some similarities and differences between the three populations
we are considering here? Although the latter may be more mature, L1 and
cL2 learners are children, that is, they fall within the window of opportunity
of the Critical Period for language. On the other hand, cL2 and aL2 learners
are similar in that they have a previously acquired language, their native
language. In contemplating whether the same underlying processes are

1
Although we saw in the previous chapter that a lot of complex constructions are not
acquired by children until much later than the age of 4.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 144 6 CHILD SECOND LANGUAGE

involved in L1, cL2, and aL2, Schwartz (2009: 66) makes the important
distinction between linguistic development and ultimate attainment in child
and adult L2ers. She argues that looking only at the end states of child and
adult L2 acquisition is not sufficient because they may have reached the
same end state via different routes. Instead, it is important to look at
language development over time. Schwartz (2009) proposes the Domain
by Age model that we shall discuss in some detail below.
To illustrate common developmental stages between cL2 and aL2, we
shall take an example from Unsworth’s (2005) work (as cited in Schwartz
2009). The Dutch language allows an operation called “scrambling,” or
movement of the definite object over negation, as (1) and (2) illustrate.

(1) Base word order in Dutch, SOV

Nijntje gaat niet de bloem plukken.
Miffy goes not the flower pick

(2) Scrambling of definite DO: DO Neg V

Nijntje gaat de bloem niet plukken.
Miffy goes the flower not pick

(3) cL2 and aL2 order: (S Aux) Neg V O

*(Nijntje gaat) niet plukken de bloem.
Miffy goes not pick the flower
‘Miffy is not going to pick the flower.’

When learning the possibility of the object scrambling over negation as in

(2), both child and adult L2 learners go through a stage that is not attested in
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

child L1 development. Under the influence of English, they produce sen-

tences as in (3), in which the object appears after the verb and negation.
Later on, they go through a stage when they produce both the attested and
the unattested order, before they converge on the target word order as in (2).
Children acquiring Dutch as a native language do not make such errors
because they are not influenced by a previously learned language.
Unsworth (2008b) compares the scrambling behavior of low-proficiency,
intermediate-proficiency, and high-proficiency children and adults L2
acquirers. The behavior that emerges is as follows: while the low proficiency
children and adults fail to scramble where they should, intermediate and
high-proficiency children and adults scramble more or less consistently.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.2 CHILD L2 ACQUISITION 145

Child L1 Child L2 Adult L2

Developmental
course

Endstate

Figure 6.1 Illustration of the Domain by Age model by Schwartz (2009)

Thus, in this experiment, participants pattern together on the basis of

proficiency, not on the basis of age. The L2 children and adults were
observed to pass through the same developmental sequence, and hence
their behavior was consistent with the claim that UG is involved in adult
L2 acquisition.
Capitalizing on such findings, Schwartz proposes the Domain by Age
model (Figure 6.1), essentially proposing that child L2 acquisition proceeds
like adult L2 acquisition in the domain of syntax (dark grey (black) arrows),
but it is like L1 acquisition in the domain of inflectional morphology (light
grey arrows). The rationale for the second part of this claim came from
work by Weerman (2002), which documents overgeneralization errors with
Dutch adjectival morphology.2 These overgeneralization errors go away by
age 6 for Dutch L1 children, but they persist for child L2 learners.
However, more recently Schwartz (2009) brings forward unpublished
data from Tran (2005), to update this latter part of the Domain by Age
model. Tran investigated whether a well-supported fact in child German is
also manifested in cL2 German. German children move the verb in second
position in the sentence only when it is finite, that is, it is correctly marked
with person, number, and tense inflectional morphology. When the verb is
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

in its infinitival form, it takes sentence-final position in child speech. The

data in (4) and (5) illustrating this contrast comes from Poeppel and
Wexler’s work on the production of Andreas, a monolingual German
acquirer at age 2;1.

(4) V2 (OVS) (Poeppel and Wexler 1993: 14, (13b))

Ein Fase hab ich
a vase have I

2
L1 and L2 children produce a schwa in contexts where there should be no schwa.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 146 6 CHILD SECOND LANGUAGE

(5) V-final (SOV) (Poeppel and Wexler 1993: 11, (11))

Thorsten Caesar haben.
Thorsten C. (=doll) to.have

Tran tested native English children learning German in Honolulu. While

the high-proficiency learners had already acquired the above contingency,
the mid and low-proficiency children had not. They placed 57% of non-finite
forms in the verb-second position in marked contrast to Andreas. Schwartz
suggests that these data support the observation that cL2 learners may get to
the desired targetlike state of knowledge, but they take different routes from
L1 children.
Additional data in this respect come from Herschehnsohn, Stevenson,
and Waltmunson (2005), who tested 6-year-old cL2 learners of Spanish in
an immersion context, a year and a half after immersion started in kinder-
garten. The researchers looked at one inflectional morpheme—the marking
of number in 3rd person subject agreement—and tested it by eliciting
answers to questions. The children were only 38% accurate on the first
testing occasion and 56% accurate on the second testing occasion, two
months later. At the same time, it is well documented that child L1 learners
of Spanish make very few commission or substitution errors3 with this
inflectional morpheme. This study, again, supports the conclusion that
cL2 development is not like cL1 development, even in the area of inflec-
tional morphology. Another recent study on cL2, Li (2012), will be dis-
cussed in Exercise 6.4. While more data and further confirmation with fresh
language pairs is always needed, the overall picture at this moment in time
remains as in Figure 6.2 (see e.g., Blom 2008, Sopata 2010).
In summary, there is accumulating evidence that cL2s are like aL2 learners
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

both in the area of syntax as well as in the area of inflectional morphology.

They go through similar developmental stages and experience difficulties with
inflectional morphology marking. A fundamental reason for this situation
may be the presence of an already learned language in the mind/brain of the
learners, that is, bilingualism itself. We shall discuss common characteristics
of cL2, aL2, and heritage speakers in the last section of this chapter.

3
A commission error is when a speaker produces a morpheme where it is not needed,
e.g., You works. An omission error is when a speaker does not produce a morpheme
where it is needed, e.g., He work. A substitution error is producing one morpheme instead
of another, e.g., He working.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.3 L 3/ L N ACQUISITION 147

Child L1 Child L2 Adult L2

Developmental
course

Endstate

Figure 6.2 Illustration of the Domain by Age model, second version

6.3 L3/Ln acquisition

Although the focus of this text is on adult second language acquisition, a

relatively new and vigorously developing area of investigation is third (L3)
language acquisition, defined as the sequential acquisition of another lan-
guage beyond a second language. In this section, when I write L3, I will
actually include additional language (Ln) acquisition as well. How is L3
acquisition relevant to L2 acquisition? Studying multilinguals can actually
help us to understand a lot about the linguistic representation of the second
language in the mind. The essential issue, and one that is hotly debated
among generative L3 acquisition researchers, is whether it is the grammar of
the first or the second (or any subsequent) language that can influence the
L3/Ln. The answer to this question is still related to the Critical Period
Hypothesis, although in a slightly roundabout way. Here is the rationale. If
the first language and the second language are equally capable or influen-
cing the acquisition of subsequent languages, then they must have equal
cognitive and epistemic status in the mind of the learner. If the learner is
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

using her entire linguistic repertoire of grammatical features, constructions,

functional morphemes, etc., then the linguistic knowledge of the second
language must be represented in the mind in a way that makes subsequent
transfer possible. If, on the other hand, the native and the second languages
are learned in a qualitatively different way and result in qualitatively
different representations, we cannot expect the L2 to have but a superficial
influence on the L3/Ln.
Researchers in this field ask whether and when “transfer” takes places in
the L3 interlanguage development. Transfer is defined as linguistic know-
ledge that can be traced back to prior linguistic experience (see definition in
Chapter 2). To take an example given in Rothman (2015), consider the case

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 148 6 CHILD SECOND LANGUAGE

of English native speakers who also speak Spanish as an L2 and are learning
Portuguese as an L3. English marks gender only on pronouns (he, she, it),
while in Spanish and Portuguese every single noun is specified as feminine or
masculine. In addition, adjectives agree with nouns for gender and number.
We call the latter “grammatical” gender. A learner of L3 Portuguese with
L2 Spanish will already be sensitive to grammatical gender being marked on
all nouns and to adjective–noun agreement, exhibiting “early knowledge of
morphological and syntactic reflexes for grammatical gender features”
(Rothman 2015: 182).
In L3 acquisition, researchers consider the potential sources of transfer
from any of the previously known languages, both in the initial state and in
subsequent development. The specific interplay between the L1, the L2, and
the L3 parameter values is being investigated. Does transfer in L3 acquisi-
tion come exclusively from the L1 as in L2 acquisition, as Leung (2006) has
proposed? Does it come exclusively from the L2 (Bardel and Falk 2007)
because it was the grammar most recently learned? Can it come from both
languages, as Flynn, Foley, and Vinnitskaya (2004) have argued, or from no
prior experience (Håkansson et al. 2002)? All of these possible sources of
transfer have been evaluated, and found support from (some) experimental
data. Linguists have also discussed whether transfer can be only facilitative
(beneficial), or it can also be harmful, so that the L3 property is not really
acquired, or is acquired with great difficulty. In this respect, the Cumulative
Enhancement model (Flynn et al. 2004), as the name itself suggests, argues
that transfer from the L1 or the L2 can be only facilitative, while Slabakova
and García Mayo (2015) show that transfer of a very frequent and salient
property can stand in the way of complete L3 acquisition. Needless to say,
language proficiency in the second language will always be an important
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

factor for transfer: after all, a learner cannot transfer what she has not really
acquired.
Another variable that appears to play a role in L3 acquisition is what is
usually referred to as the typological proximity of the languages. Let’s take
for example Spanish and Portuguese, which are closely related languages
belonging to the Romance language family; they can be considered typolo-
gically closer to each other than to English, a Germanic language. The idea
of the Typological Primacy model (Rothman 2011, 2015), based on
Kellerman (1983), is that the learner will transfer properties from the
grammar, be it L1 or L2, which she perceives to be typologically closer to
the L3. This typological relation may only be perceived, not real.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.3 L 3/ L N ACQUISITION 149

Finally, another issue debated in generative L3 acquisition research is

whether transfer happens “wholesale” at the initial state, or whether it
happens property by property and even feature by feature, from the initial
state and into the development process. The Typological Primacy model
(Rothman 2015) argues for the former view, while the Scalpel model (Sla-
bakova 2015), defends the latter view. At present, there is no clear evidence
that can answer this question definitively. Future research aiming to tease
these positions apart should involve a longitudinal study of a third lan-
guage, studied in adulthood ab initio, where the L2 and the L3, or the L1
and the L3, are typologically similar.
With so many variables to control, research on L3 acquisition involves
complicated research designs. I shall look at three studies to give the
reader a taste of the variety of language combinations studied. Flynn
et al. (2004) is a pioneering study, one of the first to demonstrate empir-
ically that the L1 is not the only source for L3 transfer at the level of
formal syntactic features and functional categories. The authors exam-
ined the production of restrictive relative clauses in L1 Kazakh/L2
Russian/L3 English speakers. Relative clauses, as we have seen in
Chapter 2, can have their head on the right or on the left of the clause.
Head directionality in Kazakh is the same as in Japanese, while Russian
and English work in similar ways. The researchers surmised that, if
transfer is always from the native language, L3 acquisition of English
by L1 speakers of Kazakh should resemble L2 acquisition of English by
L1 speakers of Japanese (there is ample literature on this acquisition for
comparison). However, the L1 Kazakh learners had an easier time
acquiring the head directionality of the English relative clauses. Flynn
et al. proposed that the L2 Russian had a facilitative effect on the acqui-
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

sition of this particular L3 construction. More generally speaking, Cumu-

lative Enhancement argues that experience in any previously acquired
language can be taken advantage of in the acquisition of any subsequent
language.
Rothman and Cabrelli Amaro (2010) investigated two L3s, French and
Italian, where English was always the L1 and successfully acquired Spanish
was always the L2. They examined properties related to the Null Subject
Parameter. French is like English in this respect, while Spanish and Italian
work similarly, that is, they are null-subject languages. For easier reference,
the configuration is given in (6), where the italicized languages work the
same way and the non-italicized languages work the other way.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 150 6 CHILD SECOND LANGUAGE

(6) L3 French
L1 English – L2 Spanish
L3 Italian

In addition, two groups of learners of French and Italian as second lan-

guages were tested. For this part of the experiment, five groups of partici-
pants judged the grammaticality (acceptability) of ten sentences in four
conditions: null and overt expletive subjects (It rains) and null and overt
pronominal subjects (He snores). The authors demonstrated that both
groups of L3 learners transferred from Spanish, which is actually the
wrong choice for L3 French. Even though English, the native language of
the learners, could have helped in the acquisition of French, the learners
were influenced by the typologically related Romance language. This find-
ing is especially clear when we compare the learners of French as an L2 and
as an L3, in Figure 6.3. These experimental results showed that transfer can
be harmful, including when the transfer is from the language perceived to be
typologically closer to the L3. In other words, they found support for the
Typological Primacy model.

10
Rejections (group average)

9
8
7
6
5
4
3
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

2
1
0
NE OE OS NS
English NSs 9.42 0 0 9.08
L2French 8 2 0.2 8
L3French 1 7.11 0.11 1.22
L2Italian 6.18 2 0.18 2.91
L3Italian 0.5 8.6 1.1 0.6
Notes: NE = null expletive subject; OE = overt expletive subject; OS = overtpronominal
subject; NS = null pronominal subject

Figure 6.3 Group average rejection of null and overt subjects in a grammaticality
judgment task, from Rothman and Cabrelli Amaro (2010)

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.3 L 3/ L N ACQUISITION 151

Before going forward, let us spend some time understanding the figure. What is
plotted, here, acceptance or rejection? What is the highest score each group of
participants could have achieved? Each column stands for a score of one
participant group in one condition; conditions are grouped together. Look at
the first column in each group. Do the English native speakers behave predict-
ably? Now look at the next two columns in each group. These are average
scores of learners of French. Are their grammaticality judgments of null and
overt pronominal and expletive subjects different? What can you say about the
learners of Italian? The crucial groups of columns to notice are the first and the
last ones, where the L3 French learners do not reject null expletive and
pronominal subjects as they should, under the influence of their L2 Spanish.

Finally, another recent study demonstrated non-facilitative transfer, too,

but from a first as well as a second language. Slabakova and Garcia Mayo
(2015) investigated knowledge of English topicalization as in (7), a con-
struction which moves the object to the front of the sentence when it is
known, or already mentioned.

(7) A: Did Janice like the wine?

B. Oh, the wine she didn’t drink (*it). She stuck to lemon ices.

(8) El libro lo compr-amos ayer.

the book ACC.CL.3M.SG buy-PAST.1.PL yesterday
‘The book, we bought yesterday.’

The languages under investigation were Basque and Spanish as L1 or L2,

English was the third language, as (9) illustrates. Again, the presence/
absence of italicization indicates similar constructions. As in Rothman
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

and Cabrelli Amaro, there was a group of native Spanish speakers learning
English.

(9) L1 Basque – L2 Spanish – L3 English

L1 Spanish – L2 Basque – L3 English

While English and Basque work similarly with respect to topicalization, the
Spanish construction4 in (8) requires a clitic pronoun lo to double the moved

4
Known as “clitic left dislocation.”

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 152 6 CHILD SECOND LANGUAGE

1
English NS
B-Sp-E (n=23) Sp-B-E (n=24) Sp-E (n=39)
(n=24)
Acceptable 5.8958 4.6804 4.1071 4.1758
Unacceptable 2.2028 4.2355 4.1792 4.4459

Figure 6.4 Mean ratings of topicalization by four groups of participants, from

Slabakova and García Mayo (2015)

object. Therefore, the unacceptable test sentences in English had a pronoun

in the object position, as in (7). The experimental design involved listening
to recorded conversations and evaluating the acceptability of test sentences
with and without resumptive pronouns, on a scale of 1 to 7. Figure 6.4 plots
the mean ratings of all the participant groups.

Look at Figure 6.4 and explain what you see. Do the native speakers exemplify
a contrast between acceptable and unacceptable topicalization? What do mean
ratings of around 4 signify in the learner groups’ performance?

Slabakova and García Mayo argued that non-facilitative transfer from

Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

Spanish is exhibited in their results, no matter whether Spanish was the first
or the second language of the trilinguals. The bilingual results also buttress
this conclusion.

Teaching relevance: “Concealed” L3 learners

It is very interesting that after researchers started paying attention to L3/Ln
acquisition, it turned out to be the case that many classroom learners are in fact
not L2 but L3 learners. For example, it is most often the case in the US college
system that students come to Portuguese or French or Italian classes with some
or a lot of Spanish L2. It is useful for teachers of such students to be aware of how
the Spanish grammar can facilitate these other Romance languages as L3s.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.4 HERITAGE LANGUAGE LEARNERS 153

In summary, the vigorously developing research in L3/Ln acquisition

focuses on the issue of the source of cross-linguistic influence. Transfer can
come from the first or the second language of the learners, and it can be
facilitative or harmful. As in generative linguistic research in general, trans-
fer is understood not as a superficial phenomenon of influence, but on a
deeper level of grammatical competence and development. Generative lan-
guage acquisition researchers do not perceive L3/Ln acquisition as a simple
extension of L2 acquisition, because of the complex sources of transfer
available to multilinguals.

6.4 Heritage language learners

We brought forward the importance of heritage language speakers for the

Critical Period debate in Chapter 4. In this section, we will expand on the
characteristics of this important population’s language knowledge.
The term heritage speaker was first introduced in the mid-1970s
(Cummins 2005) but has been gaining ground in the acquisition literature
since the 1990s. Broadly defined, heritage speakers are child and adult
members of a linguistic minority who grew up exposed to both their home
language and the majority language. For some researchers, this definition
also includes indigenous languages, not just immigrant languages
(Fishman 2006). Spanish, South Asian, Russian, East Asian, and Arabic
speakers in the US who are the so-called second generation immigrants
are representative examples. While the parents are either monolingual or
dominant in their native language, the children grow up in homes where
both the majority language and the native language are spoken. Very
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

commonly, the heritage language is the individual’s native, or consecu-

tively first language, learned in a naturalistic setting in a family environ-
ment. The heritage language could also be one of two simultaneously
acquired L1s. However, the dominant language of those individuals is
the majority language of the country and the community, exerting educa-
tional and social pressures over the weaker, heritage language. Once the
majority language is introduced, the heritage language becomes restricted
to family-centered uses and situations. Figure 6.5 and 6.6 from Montrul
(2012) illustrate the typical development of heritage and L2 languages,
where the length of the columns is meant to represent proficiency with
respect to adult norms.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 154 6 CHILD SECOND LANGUAGE

L1 = Native language (majority language)

L2 = Second language

L1

L1
L1 L2
L1 L2

Early Middle–late Adolescence Adulthood

childhood childhood

Figure 6.5 Typical development of a first (L1) and second (L2) language (after
puberty) in a majority language context, after Montrul (2012)

L1 = Heritage Language
L2 = English (in the US)
L2

L2

L1 L2 L1
L1
L2 L1
early middle-late adolescence adulthood
childhood childhood

Figure 6.6 Typical development of a heritage language (L1) in a majority language

context, after Montrul (2012)

Who exactly falls into this rather heterogeneous group of heritage

Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

speakers? According to Montrul (2011b), this group may include:

(a) simultaneous bilinguals, those exposed to the heritage and the majority lan-
guage before the age of 3–4; (b) sequential bilinguals or child L2 learners, those
exposed to the heritage language at home until age 4–5 and to the majority
language once they start preschool; and (c) late child L2 learners, children mono-
lingual in the heritage language, who received some elementary schooling in
their home country and immigrated around ages 7–8. (Montrul (2011b: 157)).

This is a good place to reinforce an important distinction in bilinguals and

multilinguals. One should be careful not to confuse language dominance
with language chronology or nativeness. Heritage speakers have acquired
their heritage language chronologically first and from birth. It is their

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.4 HERITAGE LANGUAGE LEARNERS 155

mother tongue. However, they later become dominant in another language,

spoken by the wider community around them. Very often they sound
nativelike in that second language. Countries with consistent streams of
immigration through the years, such as Canada, the UK, and the US are full
of such people, the so-called second-generation immigrants (Silva-Corvalán
1994). Teachers of foreign languages in such countries encounter many
individuals fitting this profile in their classrooms. When tested in their
weaker, non-dominant language, these speakers may be more or less profi-
cient. Indeed, according to some sources, about 30% of undergraduate
students in American colleges and universities may be heritage speakers of
an immigrant language; in California, this percentage may be even higher,
at around 40% (Carreira and Kagan 2011).

Teaching relevance: Dominance versus chronology

Identifying specific areas of linguistic knowledge in which the heritage lan-
guage speakers and L2 learners may or may not differ informs materials
development, classroom-based instructional intervention, or language pro-
gram direction. The educational needs of these two populations are just not
the same. It is important for teachers to be aware of just how different heritage
speakers are from adult L2 learners.

Which aspects of language suffer when the consecutively first language

gets used less and less over the years, until it becomes a weaker language of a
bilingual? Recent research into the competence of these speakers has
uncovered significant gaps in their linguistic competence, making them
very different from age-matched monolingual speakers.5 In general, adult
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

heritage speakers are better at aural comprehension than oral production.

This is not surprising if we keep in mind that they continue to overhear the
heritage language in their family circles although they may not speak it on a
regular basis. Furthermore, their written language skills are worse, com-
pared to adult L2 learners, because they typically do not have the benefit of
schooling in the heritage language.

5
See Benmamoun, Montrul and Polinsky (2013a,b) for a recent overview and lin-
guistic treatment.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 156 6 CHILD SECOND LANGUAGE

An important early series of studies, Au et al. (2002) and Knightly et al.

(2003), compared beginning L2 learners of Spanish to Spanish native
speakers with only a receptive knowledge of the language. They labelled
the latter speakers “overhearers.” The researchers found that the heritage
speakers were significantly more nativelike on the phonetics/phonology and
pronunciation measures, while their performance on morphosyntactic
measures, around 60% accurate, was similar to the L2 performance. Advan-
tages of heritage speakers on phonetic discrimination and pronunciation
have been confirmed for various heritage languages: Saadah (2011) for
Arabic, Lukyanchenko and Gor (2011) for Russian, Chang et al. (2008)
for Mandarin Chinese.
Although heritage speakers enjoy some advantages in phonetics/phon-
ology, their linguistic systems share many similarities with adult L2
speakers. Many of the problem areas typical of L2 learners, such as inflec-
tional morphology, complex syntax, syntax–semantics mismatches, and
discourse-related meanings seem to be problematic for heritage speakers
as well. We will illustrate the parallels between heritage and adult L2 learner
grammars with some concrete studies below.
One representative study of heritage speaker case morphology, Song,
O’Grady, Cho, and Lee (1997), investigated the ability of monolingual
Korean and heritage Korean children to use case marking for distin-
guishing the doer of the action in a transitive situation. In Korean, the
order of the object and the subject can be reversed without any add-
itional changes, because the arguments are marked with Nominative and
Accusative case. Participants heard a scrambled Korean sentence as in
(10) (similar to the scrambling in Dutch we saw earlier). They had to
choose one of two transitive construals (who is hugging who), as illus-
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

trated in Figure 6.7.

(10) oli-lul thokki-ka anacwue OSV

duck-ACC rabbit-NOM hug
‘It is the duck that the rabbit is hugging.’

Whereas the monolingual children responded above chance (50%) by age

4, even the 8-year-old heritage children did not reach chance level on these
structures. Since Korean case marking depends on four interrelated
factors—grammatical function, focus, animacy, and definiteness—the
researchers concluded that the input the heritage speakers received was

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.4 HERITAGE LANGUAGE LEARNERS 157

Figure 6.7 Sample picture used to test the attentiveness to case in the interpretation of
sentence (10) by Korean monolingual and heritage children, from O’Grady et al. (2011)
Reproduced with permission

not sufficient for them to establish a difficult form-to-meaning mapping of

this sort.
In another recent study, Albirini, Benmamoun, and Saadah (2011) inves-
tigated the narrative production of Arabic-speaking heritage speakers of the
Egyptian and Palestinian dialects. They attested quite a high level of profi-
ciency among these speakers. Still, some significant gaps in their knowledge
became visible, in comparison to monolinguals. They had problems with
agreement and tense morphology. For example, where inflected verbal
forms were expected in the narratives, these speakers frequently substituted
a simpler participial form. The researchers contend that the heritage
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

speakers use this participial form as default, when access to the verbal
paradigm fails them. Transfer from English word order was also attested.
To exemplify how heritage speakers deal with complex syntax, we shall
mention the findings of Polinsky (2011) on comprehension of relative
clauses in heritage Russian. Polinsky tested prepubescent heritage speakers,
age-matched monolingual Russian children, and adult Russian speakers.
The child heritage speakers performed on a par with the monolingual
children. Meanwhile, the adult heritage speakers had significant problems
with relative clauses. A very interesting feature of their linguistic compe-
tence was uncovered: they performed at chance on object relative clauses
(The dog that the girl saw ____) but were close to the other experimental

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 158 6 CHILD SECOND LANGUAGE

groups in their comprehension of subject relatives (The girl who _____ saw
the dog). Polinsky argued that the nativelike grammatical knowledge of
relative clauses that the child heritage speakers showed had been reanalyzed
by the adult heritage speakers into a new system allowing only subjects, but
not objects to be heads of relative clauses.
The study of heritage speakers’ competence, just as the study of cL2A, has
been developing very vigorously in the last ten years. Three major explan-
ations have been proposed for the linguistic development of heritage
speakers. Of course, all such explanations have to refer to differences
between heritage speakers and monolinguals, as well as to the similarities
between heritage speakers and adult L2 learners. One line of explanation
proposed by Montrul argues that heritage speakers exhibit signs of incom-
plete development (Montrul 2008). The reduced situational, family-based
usage can cause “arrested development” in the heritage language or lead to
incomplete acquisition of some constructions. Another explanation, pro-
posed by Polinsky maintains that heritage languages have in fact been
acquired completely, but have subsequently reduced by language attrition
(loss of native language, see the following section) (Polinsky 2011, among
others). A third explanation, argued for by Sorace and Rothman separately,
directly relates heritage language competence to reduced input, not only
quantitatively but qualitatively. Heritage speakers receive input primarily
from speakers—first-generation immigrants and other second-generation
speakers—who are themselves living in a language contact situation and
hence may be consistently providing them with linguistically changed input.
Thus, individual attrition may be reinforced through intergenerational
attrition (Sorace 2004, 2012, Rothman 2007, Pascual y Cabo and Rothman
2012). Of course, it is likely that all of these factors work in consort to some
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

extent. The common thread running through these three explanations is that
the linguistic input, variable both in quantity and in quality, and the reduced
language use, are the chief reasons why heritage grammars are different
from monolingual grammars. This is true even though heritage language
speakers are native speakers.

6.5 Language attrition

The term language attrition refers to native language regression or loss. The
speakers of such a reduced language are dubbed attrited speakers. Attrition

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.5 LANGUAGE ATTRITION 159

research examines the potential erosion of native language competence after

long exposure to another language. Although they are related, heritage
language competence and language attrition have to be clearly delineated.
In the case of attrition, a given property of language has been fully acquired
by a speaker who has typically reached a stable native grammar state, but is
subsequently lost due to reduced exposure to the native language. Attrition
is much easier to demonstrate in adults than in children: the latter may or
may not have acquired the linguistic property completely in the first place.
Recall from the previous section that Polinsky (2011) argued for attrition
being the cause of heritage speaker differences by comparing heritage
children who exhibited knowledge of object relatives with heritage adults,
who didn’t. Only this type of research design can rule out incomplete
acquisition and support attrition as the sole cause of the grammatical
deviation from native norms.
In cases of native language attrition, exposure to an L2 is often sustained
and intense and may be accompanied by a sharp decrease in exposure to the
L1. The typical attrited speaker is the first-generation immigrant, who may
be the parent of heritage speakers. The definition of attrited speakers,
similarly to heritage language, encompasses attriters with a wide variety of
linguistic situations and profiles (Schmid 2011). Under a broad definition,
attrition can be studied in balanced simultaneous bilingual or childhood L2
learners, adults, as well as heritage speaker bilinguals. A narrower definition
focuses on loss of (some parts of) the language faculty after the putative
critical period.
Language attrition is typically minimal if attriters lost contact with their
native language in adulthood. For example, Schmid (2002) found that,
after more than 30 years of living in the US, German Jews exhibited some
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

transfer from English but made very few morphosyntactic errors that could
be attributed to L1 attrition. No adult speaker, who comes into contact with
a second language and starts living in a new environment, is likely to forget
the verb conjugations, the native sound contrasts, and how to ask questions
in their native language (Keijzer 2007). But less dramatic, although still
measurable, changes in native grammars may be documented when poten-
tial attriters are compared to recent arrivals or native speakers in the
country of origin (Gürel and Yılmaz 2013, Sorace 2005).
However, not all cases of attrition are relatively mild and superficial.
Iverson (2012) discusses a case study of extreme attrition of native Spanish
in contact with Brazilian Portuguese. The bilingual speaker, called Pablo

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 160 6 CHILD SECOND LANGUAGE

(not his real name), was tested in his fifties. He was born in Chile, where he
went to school until age 13, then worked in Chile and in other Spanish-
speaking countries until he moved to Brazil in his early twenties. At the time
of the testing, Pablo had spent 30 years in Brazil with minimal contact with
Spanish. He never visited Chile nor did he speak to family members there on
a regular basis. He is a street artist, married to a Brazilian Portuguese-
speaking woman and is the father of a Brazilian Portuguese-speaking
daughter. He rarely encounters other Spanish speakers.
Iverson obtained spontaneous speech samples which demonstrate that
Pablo intersperses many Brazilian words in his Spanish narrative, not only
nouns, but verbs as well. Sometimes he produces words that are neither
Spanish nor Brazilian Portuguese but amalgamations of both, such as enton
‘then’ from the Brazilian então and the Spanish entonces. His speech, which
he considers Chilean Spanish, is not recognized as such by native speakers.
Iverson further tested Pablo, as well as two control groups of monolingual
Brazilian Portuguese and Chilean Spanish speakers of the same socio-
economic status as Pablo, and speaking the same dialects. A range of
tasks probed judgments of verb–subject word order, overt pronoun use in
discourse-neutral contexts, null objects in complex syntactic contexts and/or
with definite antecedents, interpretation of embedded overt subject pro-
nouns in ambiguous contexts, and relative clause attachment in ambiguous
contexts. Although Spanish and Brazilian Portuguese are closely related
Romance languages, the tested areas represent contrasts between those
grammatical systems. The findings were extremely robust: Pablo’s perform-
ance diverged dramatically from the Spanish control group on the whole
range of the tested constructions. He consistently performed qualitatively
(and often quantitatively) like the Brazilian Portuguese control group.
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

Iverson’s conclusion is that the whole of Pablo’s native Spanish grammar

may have restructured under the influence of the second language grammar
in the circumstances of extreme cessation of native input and usage.
In summary, we speak of language attrition to describe changes in the
native grammar when immigrants stop using their native language and carry
out most of their communication in the second language, the language of
their new linguistic community. There could be personal and societal pres-
sures for such disuse. Research to date has shown that the changes to the
native grammar, in most cases, remain superficial and are manifested in loss
of processing accuracy and speed, as well as labored access to lexical items
and complex constructions. However, extreme cases such as Pablo’s are also

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.6 COMMONALITIES AND DIFFERENCES 161

attested. Future research should address the factors making possible and
accelerating such restructuring of a native grammar through disuse.

6.6 Commonalities and differences between

the four acquisition contexts

In Chapters 5 and 6, we have been describing various types of bilingual

situations, as well as the effects of the first language on the second and the
effects of the second language on the first. In Chapter 5 we discussed
bilinguals who have two first languages (2L1), or simultaneous bilinguals.
In Chapter 6 we discussed child second language (cL2) and adult second
language (aL2) learners, multilinguals, heritage native speakers, and attrited
native speakers. Now is a good time to take stock and identify which are the
central and decisive factors affecting all of these language development
processes. We shall consider Age and Input as crucial impacts, admittedly
simplifying matters and summarizing a lot of studies for the sake of gaining
some overarching insight into bilingual language use and maintenance.

6.6.1 Is age the crucial factor in bilingual acquisition?

The short answer is no. Evidence supporting this negative answer comes
from three relatively well-established facts. One fact is that even if a child
starts acquiring two languages from birth, one of them is likely to be the
weaker language as a function of relative hours of exposure and hours of
use. There are many indications that the mere fact of becoming a (proficient)
bilingual has ramifications for both linguistic systems. For example, Flege
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

(1987) demonstrated that bilinguals pronounce some phonemes with some

intermediate value in both their languages, a value that is not attested in
either language. Dussias (2004) found a similar bidirectional interference
effect with respect to grammatical processing strategies. Such findings sug-
gest that the monolingual norm may be something that bilinguals can never
fully attain, not necessarily because they have reached the limit of their
acquisitional potential but simply by virtue of being bilingual.
A second fact militating for a negative answer to Age as the crucial factor
is that child and adult native speakers follow the same developmental paths
and make similar developmental errors. As Schwartz and others have
argued, if Age were the crucial factor, we would expect to see children and

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 162 6 CHILD SECOND LANGUAGE

adults diverge in the stages of acquisition they exhibit and the type and
rough percentages of errors they make, but they don’t.
A third factor is possible ultimate attainment in some areas of the
grammar, even if one starts in adulthood. Although it is true that not all
bilinguals become near-native speakers, it is possible for many to become
such speakers in the areas of syntax, semantics, or pragmatics. The level of
proficiency attained is usually a better predictor of truly nativelike behavior
than age of acquisition (Wartenburger et al. 2003). Although age and pro-
ficiency may be very difficult to disentangle, and they are seldom controlled
for, there are studies which point to the possibility for highly proficient L2
speakers who started learning the language across a range of ages to apply
the same processing mechanisms as natives (Herschensohn 2007). To give
just one example among many, Montrul and Slabakova (2003) investigated
knowledge of aspectual semantics in very advanced-proficiency bilinguals
and identified around 30% of them who performed within native speaker
norms. If age is not crucial for the successful acquisition of such properties,
then the critical period claim is too strong indeed.

6.6.2 The Critical Period Hypothesis and the importance of the input

If age is not the absolutely crucial factor, then what is? Some of the points
discussed in the previous sections indicate that actually input is the crucial
factor. To elaborate, simultaneous bilinguals are exposed to two languages
from birth, but the effects of bilingualism are mitigated by the amount of
time they spend using the weaker and the more dominant language. Heri-
tage speakers are also exposed to the family language from birth, but the
disuse of that language in later years in favor of the majority societal
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

language leads to dramatic reduction of fluency and accuracy, and in

many cases even to restructuring of the native grammar. Finally, attrited
speakers who left their native country in adulthood, and thus interrupted
their native language flow of communication, also lose some fluency and
experience lexical and constructional access difficulties. In extreme cases,
they may undergo changes in underlying mental representations, not just
processing preferences. The common thread between these three different
populations: simultaneous bilinguals, heritage speakers, attriters, is the
loss of usage and reduced availability of native language input. When
input is lost or diminished, there are consequences for the nativeness of the
grammar.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.6 COMMONALITIES AND DIFFERENCES 163

6.6.3 What kind of input?

Is any input going to be sufficient for language maintenance? The short

answer is that, for a language acquired to completeness to be maintained,
the input has to be copious, sustained, and continuous. Communication
(give and take) in that language is essential. Communication has to be on
diverse topics, allowing the language users to comprehend and produce a
variety of constructions of different complexity. Reading in the language,
not just oral communication, is also considered beneficial for language
maintenance, because there are different constructions and aspects of lan-
guage exemplified in oral and written varieties. For example, it is a well-
known fact (and easily checked) that heritage speakers tend to be good at
aspects of language that are reinforced in family communication, such as
household, food, and vocabulary related to everyday life. They have been
exposed to the lower frequency, academic-oriented vocabulary, and the
complex structures of the more formal registers in their second language
in school settings, and they may not have learned that more specialized
vocabulary or registers in their native language.

6.6.4 Which areas of the grammar suffer with reduced input?

The answer to this question comes from looking at the performance of

bilingual speakers in reduced input conditions on a variety of linguistic
properties. Benmamoun, Montrul, and Polinsky (2013a,b) cite morph-
ology, complex syntax, semantics, and discourse pragmatics as areas of
the grammar where heritage speakers display non nativelike characteris-
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

tics. For example, Montrul and Ionin (2010) identified non targetlike
interpretations of the definite article and the bare noun in generic contexts.
Heritage speakers accepted sentences such as *Leones son peligrosos,
‘Lions are dangerous,’ which is good in English but unacceptable in
Spanish. Heritage speakers have been shown to omit case marking or
misuse it (Montrul, Bhatt, and Bhatia 2012, O’Grady et al. 2011).
Polinsky (2006) showed that Russian heritage speakers who cannot write
in Russian had reanalyzed the gender system to contain two genders
instead of the monolingual Russian three. Kim, Montrul, and Yoon
(2009) studied the interpretation of three local and long-distance reflexives
by Korean heritage speakers. They found that while the monolinguals had

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 164 6 CHILD SECOND LANGUAGE

a three-way anaphor system, the Korean heritage speakers had a simpli-

fied system with only two pronouns.
In sum, studies have uncovered morphological instability, high variabil-
ity, and nonnative knowledge of definiteness marking, gender agreement,
case marking, lexical aspect, grammatical aspect, mood, and inflected
infinitives. As this already sizable body of data indicates, native speakers
in circumstances of reduced input demonstrate nonconvergence, simplifica-
tion, and/or reanalysis of various areas of their native grammar. Frequently,
these have been shown to be the effect of the dominant second language. It
appears that all areas of the grammar can be affected by reduced input, with
the possible exception of phonetics/phonology. However, it is still an
unanswered question why some properties within an area of the grammar
are affected while others are not. For example, Håkansson (1995) found
that Swedish heritage speakers did not have problems with the Verb Second
rule, a complex grammatical rule. Furthermore, while Russian heritage
speakers have severe problems with gender and case, Spanish heritage
speakers typically have fewer problems with these grammatical morphemes.
Theoretical answers to these questions should include an array of experience
factors affecting linguistic knowledge, as well as principled proposals based
on linguistic theory.

6.6.5 Are first and second language acquisition qualitatively different?

The short answer is no, first and second language acquisition are not
qualitatively different. If the Language Acquisition Device were perman-
ently altered after the acquisition of the native language in such a way that
subsequent languages were acquired in a qualitatively different way, then
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

the acquisition of the dominant language of heritage speakers cannot be

explained adequately (see Figures 6.3 and 6.4). Furthermore, if we find, as
we do, that even adult L2 learners attain similar proficiency to monolinguals
in some areas of the grammar, then the same language acquisition mechan-
isms must be at work. It is as simple as that: if language acquisition
processes (L1A, L2A, L3/LnA, heritage L1A, heritage L2A) start at differ-
ent ages but lead to comparable attainment levels, the mechanisms cannot
be qualitatively different.
This interpretation of the L1–L2A fundamental difference remains con-
troversial. Proponents of a strong version of the Critical Period Hypothesis
argue that it is a special cognitive-based and measurable aptitude that

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.6 COMMONALITIES AND DIFFERENCES 165

makes some but not all individuals better language learners, leading to
performances quantitatively comparable to monolinguals (Long, DeKey-
ser, Meisel). According to this line of reasoning, even if they are better
language learners, these successful individuals do not learn language in the
way L1 children do, because they rely on general cognitive skills, as well as
metalinguistic patterns of observation and explicit explanation. This type of
argument is only countered by demonstrating that successful L2 learners
have mastered properties that are not taught in language classrooms and
that are not easy to acquire based on observation of natural language input.
Such research findings exist, and we will see them discussed at length in
Chapter 10. Suffice it to mention just one study here, Slabakova (2003),
which shows that successful interpretation of taught and untaught semantic
contrasts is demonstrated by Bulgarian learners of English even at inter-
mediate levels of proficiency.
At the same time, if we maintain that L1 and L2 acquisition are funda-
mentally similar processes, we must explain the wide discrepancy in attain-
ment levels for the two types of acquisition. While a typically developing first
language is always acquired to completeness, a typically developing second
language is not guaranteed 100% success. Could the solution lie in the fact
that acquiring a second language creates a bilingual individual? Bilinguals
experience complex demands on their language faculty because they
must constantly negotiate two competing systems at the same time, both in
production and comprehension. Even if they are able to establish fully
targetlike underlying knowledge of the L2 grammar, they may fail to apply
it deterministically due to the competing linguistic systems and the height-
ened demand on processing resources. In order to use her weaker language, a
bilingual has to expend a great deal of effort at inhibiting the deeply
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

entrenched routines used by her dominant and stronger language. This

inhibition can sometimes fail, allowing the output from underlyingly intact
rules to show influence from the dominant language.
It stands to reason that all types of bilinguals (consecutive, sequential,
and adult) should show evidence of such “bilingualism effects.” In other
words, we should find more widely variable outcomes with all types of
acquisition leading to bilingualism but no such variation with monolingual
acquisition (barring impairments and traumas, of course). Interestingly,
such evidence comes from the area of simultaneous bilingualism. “Raising
children to speak a single language has a 100% success rate except in some
cases of impairment. Raising children to speak two languages only has a

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 166 6 CHILD SECOND LANGUAGE

75% success rate.” (De Houwer, 2007: 421). Why would that be the case,
even with consistent exposure to two languages from birth? We can also
construe the heritage language findings discussed in this chapter in this light:
while competence in one language goes up, competence in the other lan-
guage might go down, see Figure 6.6.6
A clinching argument comes from a recent study by Hopp and Schmid
(2013). The idea was to compare the performance of late L2 learners against
that of other, similarly fluent and proficient bilinguals, who share the effects
of cross-linguistic influence but who have acquired the same language
from birth. The two researchers compared data that they had collected
separately7 on some comparable measures. Hopp’s body of data was from
near-native speakers of German with English and Dutch as their native
languages; Schmid’s was from German attriters in Dutch and English
environments. Since the two populations took a C-test8 measuring profi-
ciency, their global proficiency could be compared directly. The ranges of
the attriters on the C-test were much larger, but the means were comparable,
and lower than the monolingual control group.
An investigation of perceived foreign accent among these groups of
speakers revealed that a minority of the L2ers (37.5%) and a majority of
the L1 attriters (72.5%) were perceived to be within the monolingual range.
However, 80% of all L2ers fell into the range delimited by the attriters.
Schmid (2013) further compares the rate of morphosyntactic errors per
minute, taken from a spoken speech sample. Although the topics and the
duration of the samples were not exactly the same, Schmid collated errors in
four categories: lexical errors, word order (syntax) errors, and two morpho-
logical errors (case and gender) (see Figure 6.8).
Let’s examine the interesting contrasts between attriters and near-natives
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

on the data in Figure 6.8. The near-native L2 speakers of German are less
accurate on the lexicon (which Schmid suggests could be due to the specific

6
The reader should not extend these arguments to suggest that there is only room for
one language in the brain and if you add another, the first will inevitably get worse. This is
not the case. All I am suggesting here is that languages suffer from loss of input.
7
Hopp (2007) and Schmid (2007).
8
A C-test is a type of language test in which learners read a brief paragraph in the
target language. The first two sentences are left intact. Thereafter, every other word is left
intact but for each alternate word, only the first half of the word is written out while the
second half is represented by a blank space for each missing letter. This is what it looks
like: every oth_ _ word i_ left int _ _ _.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.6 COMMONALITIES AND DIFFERENCES 167

1.2

Errors per minute

0.8

0.6

0.4

0.2

0
Lexical Case Gender Word order
Monolinguals 0.01 0.04 0.05 0.04
Attriters-Dutch 0.21 0.11 0.11 0.2
Attriters-English 0.28 0.11 0.06 0.2
Near-natives Dutch 1.08 0.42 0.96 0.06
Near-natives English 0.66 0.08 0.29 0.12

Figure 6.8 Errors per minute in the production of four experimental groups, from
Schmid (2013)

speech topic), but they are extremely accurate on word order, even more so
than the attriters. With respect to case marking, the English native speakers
are again out performing the German attriters. Gender-marking, on the
other hand, which is relatively unproblematic for the attriters, seems to be
hard for the near-natives. This is particularly prominent for the Dutch-
native learners of German, although Dutch has a system of gender-
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

marking that is quite similar to German. We shall look at the acquisition

of grammatical gender later (in Chapter 7), suffice it to say here that it is
considered one of the hardest functional categories to acquire due to the
high level of lexical learning involved. The overall results of this interest-
ing and rare comparison suggest that it is possible for L2 learners to
become as successful as bilingual natives on many linguistic measures,
while monolinguals outperform both, at least at the group level. As the
research design keeps constant the effect of managing two linguistic
systems online, both the similarities and the differences between the two
populations must be due to the underlying competence or the performance
of the two groups.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 168 6 CHILD SECOND LANGUAGE

In sum, the position emphasized in this chapter is that continued diverse

and comprehensible linguistic input and rich communicative use emerge as
the indispensable conditions for successful language acquisition, overriding
potential critical periods (at least in some modules of the grammar). At the
same time, Universal Grammar principles and parameters constrain learn-
ers’ hypotheses and can facilitate subsequent language acquisition. Bilin-
gualism or multilingualism effects, such as the necessity to manage two or
more competing grammar systems in production and comprehension, may
go a long way towards explaining the higher variability in additional
language attainment (no matter whether that is from birth, in childhood,
or in adulthood). It could be the case that the language aptitude some
researchers correlate with attainment is directly related to the ability to
navigate two competing grammatical systems, although this remains an
empirical question. I have argued that the null hypothesis of L2 acquisition
is that it is fundamentally similar to L1 acquisition,9 because both are
processes of human language acquisition.

6.7 Exercises

Exercise 6.1. Montrul et al. (2008) revisited the question originally posed
by Au et al. (2002): whether early exposure to the language confers an
advantage in linguistic ability to heritage speakers (n = 42) over L2 learners
(n = 44). This study investigated syntactic knowledge of gender agreement
in three tasks: a comprehension task, a written morphology recognition
task, and an oral production task. While the native speakers performed at
ceiling on all tasks (with almost 100% accuracy), the two experimental
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

groups were significantly less accurate. Yet, the results also revealed inter-
esting task effects. Within the L2 group, speakers were significantly more
accurate on the two written tasks (M = 89.5 and M = 88.5) than they were
on the oral production task (M = 72.1). The heritage speakers, by contrast,
were more accurate on the oral task (M = 89.7) than on the two written
tasks (M = 84.6 and M = 83.3). Oral versus written accuracies are signifi-
cantly different within groups. Furthermore, comparing between groups, L2
learners were more accurate than heritage speakers on the two written tasks,

9
See also Hopp (2007).

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 6.7 EXERCISES 169

while the heritage speakers were more accurate on the oral task than the L2
learners.
Discuss whether these task effects change substantially, or change at all,
the interpretation that we have espoused in this chapter: that heritage
speakers’ competence is comparable to L2 competence, due to their reduced
language input and language use.

Exercise 6.2. Find the study from Exercise 6.1 in your local library or online,
read it and discuss whether the interpretation of the authors coincides with
the interpretation stated above. Discuss whether written or oral tasks better
capture implicit linguistic competence. What do the overall results tell us
about the linguistic competence of these two groups on the property under
investigation?
Exercise 6.3. Still on the same study, consider the Figures 6.2, 6.4, and 6.7.
In which group is there more variability? How is that captured in the
graphs? Now look at Table 10. How were the individual results calculated?
Discuss whether the individual results support the group findings and your
comments on variability within groups.

Exercise 6.4. A recent longitudinal study of the initial state in child L2

acquisition, Li (2012) studies six 7- to 9-year-old Chinese-speaking children
learning English in an immersion context in the US. The children had
resided in the US between four and five months at the beginning of the
study. The school instruction and environment was in English only. The
children’s spontaneous productions were recorded once a month, for eight
months in interactions with the researcher. There was not much develop-
ment in the children’s data over that time. The findings of the study are
summarized in Table 6.1. Note that commission errors (producing a wrong
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

morpheme) on tense and agreement are around 13%.

Lardiere (1998) describes the production of the adult L2 learner Patty.
Find the article, print it out, and read it. Answer the following questions:

Table 6.1 Percentage of correct usage in the spontaneous production of six English-
learning children

Correct use of verb inflections Syntactic properties

3PSG Regular Irregular Copula Overt Nominative

-s -ed past be Subjects Subjects
16% 13% 38% 93% ~100% ~100%

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.
 170 6 CHILD SECOND LANGUAGE

Question 1: Compare the personal characteristics of Patty and the children

in Li’s study. Then compare the rates of correct suppliance for Patty and the
children on the properties listed in Table 6.1. Make a combined table. Do
you see a similar pattern?
Question 2: How does Lardiere explain the discrepancy between the accur-
acy reflected in the first three columns in Table 6.1, and in the last two
columns? Do you think that Lardiere’s explanation of Patty’s performance
can extend to the children’s performance?
Question 3: Lardiere describes Patty as a fossilized learner. Why? Do you
think that the children in Li’s study are fossilized learners, and why?
Exercise 6.5. In Section 6.3, we discussed various models attempting to
describe the L3/Ln acquisition process. Looking at the experimental designs
described in this section, provide experimental configurations of languages
and properties to test the following hypotheses:

Hypothesis A: Transfer can come only from the L1;

Hypothesis B: Transfer can come only from the L2.

What findings would support hypothesis A? What findings would support

hypothesis B?
Copyright © 2016. Oxford University Press, Incorporated. All rights reserved.

Slabakova, R. (2016). Second language acquisition. Oxford University Press, Incorporated.

Created from ncl on 2024-12-20 16:58:26.