8.

LANGUAGE
ATTRITION
Monika S. Schmid and
Kees de Bot

April 15, 2018

MARNELIE B. BLANCES
Reporter
It is often considered to be a
reversal of language
acquisition.
(fairly uncontroversial
definition)
Where language acquisition
is a process during which
the proficiency in a first or
second language increases,
in the process of language
attrition, lack of contact
leads to a reduced level of
proficiency in the attriting
language.
The task of the study of
language attrition:
• To provide a more detailed analysis of
this rather idealized picture,
• To describe the observed process of
loss from linguistic as well as
sociolinguistic perspectives, and
• To try and model the (contact)
variety of the attriting language
within given theoretical frameworks.
Analysis
• Take into account observed differences in
application of rules of grammar and lexical
selection between attrited and non-attrited
language use (i.e. what are commonly perceived
as “mistakes”)
• But, it should also attempt to describe the
linguistic behavior of attriters and non-attriters
from a more holistic perspective.
• Therefore include aspects of the attriting
language even where it is not “deviant” in an
immediately obvious way
• Establishing factors such as type-token frequency,
lexical richness, or grammatical complexity.
 Study that focuses
merely on “what is If these strategies are
perfected in a simplification
lost” of linguistic system,
• Fails into account • Her speech might
avoidance strategies very well show up
that she might have little or no
developed in order “interferences’ at
all, and the
to deal with her
emerging picture
reduced capabilities. might be skewed if
“deviant”
utterances are all
that is considered.
 The picture of the attrited language
which thus emerges should help us
understand
• how different linguistic levels are affected by
the attritional process,
• How different sociolinguistic variables affect
the attritional process, and
• Whether any of the theoretical models
available can account for these observations.
8.2 Models and Theories
• 4 theoretical models and frameworks
1. Jakobson’s regression analysis
2. Language contact and language
change
3. Universal Grammar and parameter
setting
4. Psycholinguistic questions of
accessibility
8.2.1 Regression
• At the center of this hypothesis is the
assumption that
• The pattern of language dissolution in the aphasics
is similar, but in reverse order, to the pattern of
language acquisition in children.
• Those aspects of language competence acquired last,
those that are most dependent on other linguistic
development, are likely to be the first to be
disrupted consequent to brain damage;
• Those aspects of language competence that are
acquired earliest and thus “independent” of later
developments are likely to be most resistant to
effects of brain damage.
 The fact that languages are
acquired in stages by children
has been taken to suggest that
language competence is
“layered”, and that attrition
will work its way from the
topmost layer to the bottom
(Andersen, 1982, p. 97;
Berko-Gleason, 1982, p. 14;
Caramazza & Zurif, 1978, p.
145; Seliger, 1991, p. 227).
A related approach is based on
the notion of frequency of
reinforcement, hypothesizing
that it is not what is learned
first but is learned best that is
least vulnerable to language
loss (Berko-Gleason, 1982, p.
21; Jordens et al., 1986, p.
161; Lambert, 1989, p.7).
The difference between these two lines of thought,
as well as the major theoretical problem in
connection with regression hypothesis, can be
reduced to the two basic competing frameworks in
the theory of L1 acquisition:
The nativist The cognitivist
(Chomskyan) (Piagetian)
• If the sequence of L1 • If the cognitive concepts
acquisition is seen as that are seen as
determined by an prerequesites for the
innate language learning acquisition of a certain
capacity developing
autonomously feature-e.g., the concept
(Chomsky, 1965, pp. of singularity and
27-37), then the plurality, which the
hypothesis that the loss child must have in order
of this autonomous to acquire the
system will proceed in singular/plural
inverse order appears at distinction-are not lost,
least possible.
 The nativist The cognitivist
(Chomskyan) (Piagetian)
• The linguistic capacity • there is no reason
is seen as being paced why the
by growth of grammatical
conceptual and features that
communicative
express them should
capacities in L1
be.
acquisition, then such
an assumption would
not make sense:
8.2.2 Language contact and
language change
• The notion that in situations of language
contact and ensuing language change the
modifications that can be observed in the
linguistic system of one of these languages
are entirely or in parts due to language’s
encroaching on other is fairly widespread
and probably true to some extent.
• Code-switching
• Code-mixing
Grammatical system:
• A clear distinction of cases of language contact
from modifications within the linguistic system
of one language that are not due to influences
from the other is often problematic.
• Studies on language death as well as situations of
intense language contact, e.g. creolization
• The distinction between externally and internally
induced linguistic change in language attrition
was made by Seliger & Vago (1991, p. 10), who
identified different strategies of linguistic change
caused by these two forces.
Studies of attrition within this
framework have to be based on a
comparison of linguistic features of
both languages, trying to isolate
phenomena that can only be due
to interlanguage effects against
mistakes that are internally
induced. In this context, the role of
contrast between the two
languages is clearly a determining
factor, but speculations as to its
effect are contradictory:
• It has been • An alternative
hypothesized that hypothesis is that a
features that are certain stage in
cognate in L1 and L2 language attrition,
are more likely to be due to lack of input in
retained while the attriting language
categories that do not (AL), the grammar of
have an equivalent in the non-attriting
the L2 will be lost language (NAL) will
both in language become a source of
attrition and “indirect positive
language death evidence” which will
(Andersen, 1982, p. affect grammatically
97; Lambert, 1989, judgments in the AL
p. 75; Sharwood (Seliger, 1991, p.
Smith, 1989, p. 193; 237).
U. Weinreich, 1953,
p.43)
 The two linguistic systems will interact
in those domains where both of them
contain a rule which serves the same
semantic function and

• “that version of the rule which formally less

complex and has a wider linguistic
distribution… will replace the more complex
and narrowly distributed rule” (Seliger,
1989, p. 173).
Difficulties in the distinction of
processes of interlanguage and
language vs. internal simplification
of a system notwithstanding, some
processes that were predicted or
have been shown to obtain in
language attrition are qualitatively
different from processes that could
be explained by interlanguage
alone.
Language loss is thus often seen
as a form of language change
that is speeded up within the
individual or within the
community.
8.2.3. Universal Grammar (UG)
• Not so much based on an observable
sequence of acquisition but on grammatical
reasons for this sequence.
The parameter view on language acquisition
and language attrition is based on Chomsky’s
notion of a UG which contains a set of fixed
principles and certain open parameters
which are set during acquisitional process
(Chomsky, 1981, p.4; Ingram, 1989, p.64;
Seliger & Vago, 1991, p.12)
The parameter view has instigated
many studies into first and second
language acquisition with a view to
establishing factors such as:
• Are children born with an innate knowledge
about universal properties of the linguistic
system?
• If a parameter is set to a specific value, can
that setting ever be neutralized (e.g., in L2
acquisition, if the settings for L2 differ from
those in L1)?
• The role of markedness in this context: Can
a marked parameter be reset to an
unmarked setting in L2 acquisition?
Within the framework of L1
attrition,
• It has been proposed • Sharwood Smith &
that this process van Buren
might involve the hypothesize that since
“unmarking” of parameter settings
parameters that have are influenced by lack
been set to a marked of evidence through
value in L1 lack of contact,
(Hakansson, 1995, p. marked values in L1
155; Sharwood might persist
Smith, 1989, p. (Sharwood Smith &
199). van Buren, 1991,
p.26). As yet, there is
no data to support
this view.
8.2.4 Psycholinguistics
• Augments the perspectives on language
internal and acquisitional factors by taking
into account features of processing and
memory retrieval, dealing with more general
psychological issues like the accessing and
forgetting of information.
• It thus reflects the growing emphasis on
psycholinguistic processes in bilingual speech
production at large that the past decade has
witnessed.
8.2.4 Psycholinguistics
• The question of whether attrition merely
affects procedural knowledge, or whether
the actual knowledge of the language can
become deteriorated (Ammerlaan, 1996,
p.10) or, whether knowledge once acquired
can ever be lost from memory-has not
conclusively been resolved, but evidence
overwhelming points toward what
difficulties there are being only temporary.
 Ever since Weinreich’s seminal study on language
contact (U. Weinreich, 1953), contact linguistics has
attempted to provide a classification of linguistic
material terms of likelihood for transfer. It is generally
agreed that there is a cline of “borrowability” within
the linguistic system; that lexical items are more easily
borrowed than functional or grammatical ones.
8.3.1 Lexicon
 the use of NAL items in AL discourse (code switching)
 specificity of meaning , e.g. what has been called
“semantic extension” (Romaine, 1989, p. 56); “semantic
transfer” (de Bot & Clyne, 1994, p. 20), or “loanshift”
(Haugen, 1953)
 an AL word which is homophonous to an NAL item
with a different meaning is used in inappropriate
contexts (Romaine, 1989, p. 56)
 zerstreuen (‘to scatter’) used with the meaning of
English destroy, appropriate German zerstoren
8.3.2 Morphology
it has been predicted that the attrited variety will exhibit an
overall reduction in morphological complexity, resulting in
a more analytical structure. Features that have been
mentioned in this respect are:
• interlanguage effects in free morphemes,
• reduction in allomorphic variation,
• loss of agreement, especially across phrase boundaries,
• a from inflectional devices and allomorphic variation
toward more regularized or analytic forms,
• a trend toward periphrastic constructions (e.g., from an
inflected future tense to a go-future),
• grammatical relations tend to be encoded less by bound
morphemes and more by lexemes.
8.3.3 Syntax
 The assumption that word order is a domain which is
vulnerable to simplification processes in language
attrition seems intuitively convincing: many languages
offer their speakers the possibility to express what they
want to say in structures with a variation in
complexity,
e.g., hypotactical structures with large number of
embedded clauses vs. straight paratactical
constructions.
Studies:
 Yagmur’s investigation of Turkish relative clauses
 Hakansson’s study of the V2 rule in Swedish
 Schmid on verbal placement in German
8.4 SOCIOLINGUISTIC FACTORS
8.4.1 Age at onset of attrition
8.4.2 Education
8.4.3 Time
8.4.4 Gender
8.4.5 Contact
8.4.6 Attitude and motivation
8.4.7 Community factors
8.4.7.1 Identity
8.4.7.2 Ethnicity
8.4.8 Data Collection
 8.5 What Is
“Language Attrition”?
1. Collect data from a monolingual control
group and test whether the language use of
(presumed) attriters and non-attriters
shows any statistically significant differences
(Andersen, 1982, p. 85; Jaspaert, Kroon &
van Hout, 1986)
2. Data from attrition studies have sometimes
been compared to data gleaned from
statistical analyses of the distribution of the
variable under investigation in “normal”
language data (e.g., Hakansson, 1995).
However, the majority of studies done on
language attrition so far do without
comparisons of this kind. Such an approach
tacitly assumes the “mistakes” that occur in
the language of attriters to be “competence
error,” while non-attrited speakers’ language
only contains “performance errors” which
“are not representative of their ordinary
language use, and which can be corrected
by them if they are asked to do so”- slips of
the tongue.
Problems in the studies:
1. Even studies which adopt a comparative perspective
often leave us with unsatisfactory picture of an
“attriter”.
2. Is that of dialectal or sociolectal variation in the L1.
3. The fact that any classification of such as
“interferences” or “mistakes” will lead a skewed
picture of some individual’s attrition process leads
directly to the next methodological problem in
research on language attrition: what is to be
counted a “mistake”?
4. The overwhelming majority of language attrition
studies have concentrated on “what is lost” to the
exclusion of “what is retained.”
 Methodological considerations:
1. ideally, a longitudinal study design in order to establish the
attriters’ proficiency at earlier stages of the attritional
process instead of a monolingual control group;
2. Comparison of the data to non-attrited speech, either from
a monolingual control group or from existing statistical
analyses of the language under investigation;
3. Exclusion of the possibility of dialectal or sociolectal variety
within group of attriters under investigation;
4. Classification of “mistakes” from the data collected by more
than one judge;
5. Establishment of a “linguistic complexity index” to augment
the data on “what is lost” by “what is retained”.
The actual findings from
language attrition studies are
nowhere near this point: there is
not a single case where any
individual made any mistake in all
possible cases, and most of the
time the overwhelming majority
of structures are correct.
THANK YOU!