Regression and Threshold Hypotheses in English Language Attrition Through Computer Aided Education: A Computer Technology Assisted Behavioral Study
Regression and threshold hypotheses are widely-held theories
in the field of language attrition despite the fact that few empirical studies
have been carried out to validate them. This study, aiming to validate both
hypotheses, examined the reaction time and error percentage of three groups
of participants who were middle school students, workers and English teachers
with the aid of computer via an empirical study. The results showed that the
threshold hypothesis was nearly sound. In contrast, the regression hypothesis
was hardly convincible. The possible reasons for the results were also explored
and further suggestions regarding language attrition were put forward as well.
Received: September 19, 2013;
Accepted: November 21, 2013;
Published: January 25, 2014
Language attrition is, in this study, defined as the loss of a first, second
or foreign language or a portion of that language over a certain incubation
time at the individual level. It is considered as language proficiency degradation
during the process of natural language contact rather than the phenomenon which
happens when learners suffer some trauma or any other pathological hurt.
There are numerous hypotheses in the field of language attrition, such as the
threshold hypothesis, the regression hypothesis, the activation threshold hypothesis
and the inverse hypothesis. The regression and threshold hypotheses are among
those that received the most debates and theoretical explorations.
The regression hypothesis is widely acknowledged, but evidenced by few empirical
studies, possibly due to the fact that many researchers consider it as self-evident
and hence no need to conduct any experiment. Regression hypothesis supposes
that language attrition is an inverse process compared to language acquisition,
i.e., first in, last out or vice versa. In other words, the processes of language
attrition and acquisition mirror each other (Jakobson, 1941).
For another, threshold hypothesis is a well-known theory in the discipline of
language attrition as well. It means that if learners language proficiency
reaches a certain level or a set of language knowledge has been systematically
internalized in the brain, then the knowledge will become relatively resistant
against attrition, i.e., it will not be easy to be attrited.
The overall aim of the study is to (1) Test the regression hypothesis and (2)
Test the threshold hypothesis. The questions raised in this study are: (1) Is
the regression hypothesis convincing? and (2) Is the threshold hypothesis convincing?
Two hypotheses were therefore established as: (1) The regression hypothesis
is convincing and (2) Threshold hypothesis is convincing.
Comparisons of different language systems in flux have traditionally been popular
in linguistic research (Keijzer, 2009), as the
study of language during its unstable or changing phases is an excellent tool
for discovering the essence of language itself (Slobin,
1977). Only when fluctuating language varieties are compared can constraints
that govern all these unstable phases be classified (Keijzer,
2009). In other words, it is when things go wrong that a window on the grammar
in the speakers mind can be provided, allowing insights into how language
functions when things are not quite right (Keijzer,
2009; Corder, 1967).Various fluctuating language
systems, such as varieties arising from historical language change, language
contact and pidginization and creolization (Slobin, 1977)
have been investigated in this tradition (Keijzer, 2009).
It is the comparison between language acquisition and language attrition which
has traditionally received most attention, since symmetry in the construction
and dissolution of language may tell us more about the structure and storage
of language (Keijzer, 2009; De
Bot and Weltens, 1991). In particular, such research has been guided by
the question whether sequences and patterns found in language attrition are
the reverse of those observed in language acquisition. This idea is captured
in the regression hypothesis (Keijzer, 2009).
Dynamic Systems Theory (DST) is worth mentioning when discussing the regression
hypothesis. Dynamic System Theory views attrition and acquisition as two sides
of the same coin: language is dynamic and every language user passes through
stages of growth and decline throughout his or her life (De
Bot, 2007). Although it is hard to predict the changes from one point to
another because of all the dynamic variables that are involved, there are certain
set points to which the language system is drawn, so-called attractor states.
As these occur in both language attrition and acquisition, similarities are
likely to show up (Keijzer, 2007). DST basically originates
in the natural sciences and mathematics. Therefore, it remains unclear if it
could be applied to the regression hypothesis in the field of language attrition.
Few studies indicated solid support for regression hypothesis though it is
widely acknowledged and applied in the research. The young were found to suffer
more attrition than the old and the productive nouns appear more easily attrited
than the receptive (Cohen, 1989). This does not conform
to the regression hypothesis. The early obtained German linguistic elements
kept longer than those got later and the last acquired knowledge is attrited
earlier than the first obtained to a large extent (Kuhberg,
1992). This implies that the acquisition of German language appears the
mirror of the attrition only partly and it is hard to infer whether the rest
can mirror each other or not. The attrition in morphology may coincide with
regression hypothesis. The syntax, however, may not conform to the regression
hypothesis at all (Keijzer, 2009).
As for the threshold hypothesis, it is only evidenced by Hanson
(1999), who confirmed it via investigating the negative English sentences.
Despite the fact that neither regression nor threshold hypothesis has been
fully supported and that few empirical studies have been conducted, most of
related literature has considered both hypotheses reasonable and both are widely
applied to the research of language attrition.
|| Developmental sequence of negative english sentence
How to validate the regression and threshold hypotheses seems a puzzling job
which may account for the phenomenon why they have rarely been evidenced. It
is reasonable to identify both hypotheses in terms of negative English sentences
since many scholars have reached a general consensus on developmental sequence
of negative English sentence except for some minor differences (Milon,
1974; Cancino et al., 1978; Wode,
1981; Larsen-Freeman and Long, 1991). There is a sound
generalization of the sequence described in Table 1 (Larsen-Freeman
and Long, 1991).
As mentioned above, the regression hypothesis in language attrition refers
to that the sequence of language acquisition mirrors that of attrition. Since,
the sequence of acquisition in negative English sentences has been universally
acknowledged, the most important task for this study is to identify the sequence
of attrition in negative English sentences. As shown in Table
1, it is commonly accepted that there are four stages in acquisition of
negative English sentences, if the attrition sequence is the opposite of the
acquisition, then the regression hypothesis may be valid. Otherwise, it may
be inappropriate to consider the regression hypothesis as a solid rule. However,
whether we can prove or disprove the opposite sequence in this study, it may
be difficult for us to determine the soundness of regression hypothesis. The
determination of sequence of negative English sentences may not decide the sequence
of other types of sentence, such as simple sentence, compound sentence and complicated
sentence. In other words, it is possible that even the sequence of negative
English sentences conforms to the regression hypothesis while other sentences
may not. After all, to decide the sequence of acquisition and attrition in negative
English sentences can at least identify the regression hypothesis in terms of
negative sentences, hence meaningful.
METHODS ETHICS STATEMENT
The research lacks consent because the data were analyzed anonymously. The
research has been approved by the Authors institutional review board-School
of Foreign Languages of Hohai University which waived the need for written informed
consent from the participants.
All the participants were Chinese speakers, for whom English was a foreign
language. Generally, participants were students from Nanjing Foreign Languages
School (the best middle school in Nanjing City), English teachers and workers
who had received undergraduate computer aided English education. Workers did
not receive any English education and had no immediate access to English when
working. Specifically, the above participants were divided into three groups.
Group 1, to identify the developmental sequence of negative English sentence
acquisition, consisted of 12 middle school students (3 from 1st-4th semester,
respectively), 6 females and 6 males (mean age: 17.5). Group 2 was made up of
12 workers who had completed their undergraduate education (3 workers had graduated
for 6, 12, 18 and 24 months, respectively), among whom 6 scored over 500 points
and 6 less than 400 points in College English Test Band 4 (mean age: 24.3).
Group 3 (mean age: 31.6) involved 12 English teachers who had taught English
for at least 1 year, was regarded as the control group to compare with the other
Stimuli consisted of 120 negative English sentences designed according to four
developmental stages summarized in Table 1. Every 30 sentences
represented one stage. Participants had to judge whether each sentence was true
or false. Each sentence was presented at a time in white at the center of a
black screen, connected to a computer. The stimuli were 1 cm high and on average
5 cm wide. The viewing distance was approximately 50 cm.
Each participant was tested individually. The instruction which appeared on
the screen at the beginning of the experiment, stressed speed as well as accuracy.
Each trial started with a black screen followed by a cue. After 1000 m sec,
the stimulus was presented in the middle of the cue frame. The cue and the stimulus
remained on the screen until a response was given. The interval between the
response of the participant and the next cue was 500 m sec.
Negative English sentences were within-subjects independent variables. Reaction
time and right percentage were measured as dependent variables. At the beginning
of the experiment, participants performed a short practice block with four trials
in order to get used to the key and to see each cue once which was followed
by an experiment block with 120 trials. Participants were required to press
1 if they thought the sentence was right and 2 if wrong.
The first two trials of each block, as well as RTs above 3500 m sec or below
200 m sec (1.5% of otherwise correct trials) were discarded from data analysis.
Hypothesis 1: Regression hypothesis is convincing: Generally, the RT
analysis of variance (ANOVA) reveals significant main effect at different stages
of development of negative English sentences for middle school students which
shows that the developmental sequences of negative English sentences are in
conformity with those demonstrated by Larsen-Freeman and Long
(1991). On the contrary, the RT analysis of ANOVA fails to reveal significant
main effect at different stages of attrition of negative English sentences for
workers which indicates that the forgetting processes of negative English sentences
do not mirror the acquisition. This result obviously does not support the regression
Specifically, mean Reaction Time (RT) of first-stage negative English sentences
of students in 1st semester appears significantly less than that of students
in other semesters (F = 7.43, p<0.05); RT of second-stage negative English
sentences of students in 2nd semester appears significantly less than that of
students in other semesters (F = 9.35, p<0.05); RT of third-stage negative
English sentences of students in 3rd semester appears significantly less than
that of students in other semesters (F = 3.31, p<0.05); RT of fourth-stage
negative English sentences of students in 4th semester appears significantly
less than that of students in other semesters (F = 7.82, p<0.05). This shows
that the students in the 1st semester master the first-stage negative sentences
better than those in other stages that the students in the 2nd semester master
the second-stage negative sentences better than those in other stages that the
students in the 3rd semester master the third-stage negative sentences better
than those in other stages that the students in the 4th semester master the
fourth-stage negative sentences better than those in other stages. In a word,
the findings as advocate the developmental sequences of negative English sentences.
Similarly, it is found that mean Error Percentage (EP) of first-stage negative
English sentences of students in 1st semester appears significantly less than
that of students in 3rd (p = 0.049) and 4th semesters (p = 0.01) while no insignificant
differences are found in 2nd semester (p = 0.576); mean EP in second-stage negative
English sentences of students in 2nd semester appears significantly less than
that of students in 1st (p = 0.09) and 4th (p = 0.04) semesters while no significant
differences are found in the 3rd semester (p>0.05); mean EP in third-stage
negative English sentences of students in 3rd semester appears significantly
less than that of students in other semesters (p<0.05); mean EP in fourth-stage
negative English sentences of students in 4th semester appears significantly
less than that of students in the 1st (p = 0.04) and 2nd (p = 0.08) semesters
while no significant differences are found in the 3rd semester (p = 0.065).
It could be summarized that the findings are generally in conformity with the
developmental process of negative English sentences.
However, we found different pictures in terms of workers who scored over 500
points in CET4. For the workers who scored over 500, after they worked for half
a year, RT in the first-stage negative sentences is significantly longer than
that in the second (p = 0.00) and third-stage sentences (p = 0.04) while insignificantly
shorter than the fourth-stage sentences (p = 0.503); after they worked for one
year, RT in the second-stage negative sentences is significantly longer than
that in the third (p = 0.00) and fourth-stage (p = 0.009) sentences while insignificantly
shorter than the first-stage sentences (p = 0.646); after they worked for one
year and a half, RT in the third-stage negative sentences is significantly longer
than that in the other sentences (p = 0.00); after they worked for two years,
RT in the fourth-stage negative sentences is insignificantly longer than that
in first (p = 0.039), second (p = 0.962) and third-stage sentences (p = 0.955).
This result demonstrates that only after the workers worked for one year and
a half, they can forget the knowledge in third-stage negative sentences much
more than that in the other stages which indicates the conformity with the regression
hypothesis. By contrast, when workers worked for other periods studied, we failed
to find any evidence supporting the regression hypothesis.
Workers who scored less than 400 points in CET4 also presented a similar picture.
RT in the first-stage negative sentences for the workers (CET<400) who worked
for half a year is significantly shorter than that in the second (p = 0.024),
third (p = 0.018) and fourth-stage (p = 0.026) negative English sentences. RT
in the second-stage negative sentences for the workers (CET<400) who worked
for one year is insignificantly longer than that in the first (p = 0.478), third
(p = 0.108) and fourth-stage (p = 0.377) negative English sentences. RT in the
third-stage negative sentences for the workers (CET<400) who worked for one
and a half years is insignificantly shorter than that in the first (p = 0.421),
second (p = 0.314) and fourth-stage (p = 0.427) negative English sentences.
RT in the fourth-stage negative sentences for the workers (CET<400) who worked
for half a year is insignificantly shorter than that in the first (p = 0.115),
second (p = 0.663) and third-stage (p = 0.646) negative English sentences. It
could be summarized that RT for the workers who scored less than 400 points
in CET4 did not show any rule in accordance with the regression hypothesis.
Furthermore, it was not found that there were significant differences between
EPs in either different stages of negative sentences or workers who worked for
different periods. Therefore, the regression hypothesis did not obtain any evidence.
Therefore, the hypothesis the regression hypothesis is convincing was rejected.
Hypothesis 2: Threshold hypothesis is convincing: It was nearly evidenced
that English teachers commanded negative English sentences significantly better
than students and workers. RTs in all the negative sentences for English teacher
were significantly shorter than that for all the students and workers except
in the fourth-stage negative sentences for workers (CET4>500) who worked
for 2 years (T = -1.892, p = 0.065). In addition, it was also revealed that
English teachers made significant less errors than other participants since
EP in them appeared significantly less than that in other participants (p<0.05).
It was also generally supported that workers who scored over 500 in CET4 retained
significantly more knowledge of negative English sentences than those who scored
less than 400. RT in the first, second and third-stage negative English sentences
of workers who scored over 500 in CET4 (worked for half a year) was significantly
shorter than that of those who scored less than 400 in CET4 (t = -3.963, p =
0.00; t = -8.051, p = 0.00; t = -18.104, p = 0.00). However, RT in the fourth-stage
negative English sentences of workers who scored over 500 in CET4 (worked for
half a year) was insignificantly longer than that of those who scored less than
400 in CET4 (t = 1.072, p = 0.290). The above data, to a large extent, supported
the threshold hypothesis. Another evidence supporting the threshold hypothesis
was that EP in workers who scored over 500 was significantly less than those
less than 400 (p<0.05). In other words, workers whose CET4 scores were over
500, showed their relatively solid knowledge in negative sentences, since the
correct rate of their choices was significantly higher than those whose CET4
scores were less than 400. On the other hand, workers with scores less than
400 performed significantly more poorly than those with scores over 500. This
revealed that those who scored over 500 points in CET4 appeared more resistant
to attrition of negative sentences while those who scored less than 400 points
seemed more vulnerable to attrition. Therefore, the hypothesis the threshold
hypothesis is convincing was accepted.
It was revealed that the developmental sequences of negative English sentences
were in conformity with those described by Larsen-Freeman
and Long (1991). Nevertheless, we only found mirror symmetry in the workers
who worked for one and a half years in terms of RT rather than EP. In other
participants, no mirror symmetries were found which indicated that the regression
hypothesis was merely partly supported.
The regression hypothesis may be selective instead of overall. It was argued
that the features of adjectival inflection (noun phrase morphology), simple
present tense, auxiliary selection (verb phrase morphology), subordination and
discontinuous word order (syntax) did not reveal any mirror symmetries between
the emigrants and adolescents and regression might be more prevalent in the
area of morphology than in syntax (Keijzer, 2009).
This study considered regression by investigating the syntax, i.e., the negative
English sentences and found no enough evidence in favor of regression. This
study is supportive of Keijzer (2009) findings.
Furthermore, there are other arguments that are not consistent with the regression
hypothesis. A review of the literature demonstrates that there is very little
evidence for this strikingly durable and seemingly sound hypothesis. For example,
Jordens et al. (1989) explored German case-marking,
because the linguistic feature meets the conditions of gradualness and a more
or less fixed order of acquisition (p. 180). They found no evidence that
the sequences of both L1 attrition and acquisition mirrored each other, although
they suggested that L2 attrition, different from L1, might conform to the regression
hypothesis. Similarly, Hakansson (1995) study of syntax
and morphology in the language of expatriate Swedes did not find any demonstration
supporting that the sequence of acquisition mirrors that of attrition.
Besides few empirical studies supporting the regression hypothesis, theoretical
reasons to validate it are also needed. When Schmid (2002)
attempted to tie the regression hypothesis to Chomskyan nativism, such that
attrition is the reversal of an innately specified 4 Kathleen Bardovi-Harlig
and Stringer (2010) sequence of autonomous linguistic development, there
is arguably a misconception of how generative linguists view both L1 and L2
acquisition Bardovi-Harlig and Stringer (2010).
The regression hypothesis, in other words, the notion last in, first
out and best learned, last out Bardovi-Harlig and Stringer
(2010) is also frustrated by the opinion of Moorcroft
and Gardner (1987): the statistical analysis of some individual grammatical
elements showed that most recently learned structures are more likely to be
affected by loss than others, suggesting that a thoroughly learned structure
is relatively immune to language loss (p. 339).
This study might have revealed one of the least consistent findings with regard
to the regression hypothesis. Only workers who worked for one and a half years
produced mirror symmetry in terms of RT. On one hand, this might be related
to the research methodology. Comparison between acquisition and attrition of
negative English sentences requires high precision when recording participants
behavior. Many factors such as mood, environment, operation of computer software
and interest in the research may have exerted some influence on the result.
On the other hand, regression in foreign language attrition may be not an overall
phenomenon but a subtle and selective one. In other words, some linguistic features,
such as morphology, may conform to the regression hypothesis while others, such
as negative English sentences, may not.
Participants different first languages might have contributed to the result.
In this study, all the participants speak the same first language--Chinese which
is significantly different from English in terms of morphology and syntax. In
syntax, for example, the negation on predicates, subjects and adverbials in
English may be placed in different positions in Chinese. To negate a concept,
there are great differences between English and Chinese, ranging from lexica
to structure. In English, many word categories can be used as negation and examples
are noun, pronoun, verb, adjective, adverb, conjunction and so forth. In Chinese,
the negation lexica are relatively limited. The negation structure in English
also varies from the whole negation, partial negation, double negation to half
negation which is presented in different forms in Chinese. Participants with
Chinese as the first language may have been interfered by Chinese negation when
joining the experiments; hence the results may have been influenced.
Another hypothesis related to the regression hypothesis is that what is the
least subject to language attrition is not the first learned knowledge but what
is deeply impressed which is an important conception stressing enforcement of
what is learned (Berko-Gleason, 1982; Jordens
et al., 1986; Lambert, 1989). In case that
a learned linguistic feature or knowledge is enforced to a certain degree, it
will be more resistant against attrition. It is referred to as threshold hypothesis,
involving both the activation threshold hypothesis on the basis of neurolinguistics
proposed by Paradis (2007) and the widely held critical
threshold hypothesis (Neisser, 1984). It was found
that the activation threshold hypothesis was applicable to aphasic patients,
since whether the linguistic representation could be reactivated was found to
be partly reliant on frequency of use prior to the damage in brains (Paradis,
2007), i.e., the higher the activation threshold is, the greater the stimuli
are needed to reactivate the representation. The critical threshold hypothesis
seems a more important hypothesis influencing L2 attrition. It was suggested
that in the process of acquisition there might be a general critical threshold
after which language knowledge became relatively resistant against attrition
Language attrition might be selective rather than global due to influence of
mother tongue. There is one well known theory in the research of language attrition
and acquisition which is referred to as flashbulb memories. Flashbulb
Memories have been defined as particularly vivid and long lasting (autobiographical)
memories for circumstances in which one first learned of a very surprising
and consequential (or emotionally arousing) event (Brown
and Kulik, 1977). Brown and Kulik (1977) emphasized
the evolutionary importance of this biological Now Print mechanism
(originally postulated by Livingston (Livingston, 1967)
that may have been crucial for survival in circumstances when one had to remember
the details of potentially life threatening events (e.g., the time and location
of first appearance of the rival tribes). In this study, the mother tongue might
have acted as Now Print in participants brain, leading to
confusion and attrition in a foreign language. Participants might have established
relatively solid memories and knowledge of some linguistic structures. This
may explain the reason why some linguistic elements were attrited while others
still remained and why some conformed to the regression hypothesis while others
The regression hypothesis could not be comprehensively discussed without exploring
the theory of threshold. According to Brown and Kuliks study, flashbulb
memories were encoded by a special brain mechanism that switched on automatically
whenever the level of stimulus or consequentiality exceeded a certain threshold.
Despite the fact that the memory trace is not an identical (photographic) copy
of the reception event, it is nevertheless fairly detailed and virtually unsusceptible
to any decay or attrition for many years. The level of participants English
proficiency might have reached the threshold because of some extra/intra-linguistic
factors, e.g., highly intense contact with English, frequent interaction with
English speakers, so much as to form an internalized brain mechanism resistant
against attrition. On the contrary, if the proficiency has some distance away
from the threshold, the linguistic elements might be susceptible to attrition
or forgetting. To cite an example, in Chinese no and not have
have the highest frequency of use of the negative adverb (Guo,
2010), but in English many alternatives, such as hardly, never, little,
few, are frequently available. Learners whose mother tongue is Chinese possibly
acquire no and not have much better than other negative
words, thus form Now Print mechanism in their brain and consequently
these printed linguistic elements are more resistant against attrition.
Anxiety might be easily aroused in the situation of language acquisition (Bailey
et al., 2000; Price, 1991) which might have
led to different results. As argued by Horwitz et al.
(1986), oral practice in the classroom is the catalyst for anxiety among
Foreign Language (FL) or Second Language (SL) learners and the main sources
for anxiety should possibly be listening and speaking activities. Language anxiety
tends to arouse problems to SL/FL learners and influence Learners cognitive
performance (Eysenck, 1979; MacIntyre
and Gardner, 1994a, b), so that their ability to
process and short-circuit the incoming language is weakened. Many studies have
explored the negative influence of anxiety on learning achievements in terms
of all four language skills: speaking (Liu, 2006), listening
(Elkhafaifi, 2005; Kim, 2000;
Mills et al., 2006; Song,
2005), reading (Argaman and Abu-Rabia, 2002) and
writing (Cheng, 2004; Klima and
Bellugi, 1966) used taped conversation as referred data. Except negative
English sentences, learners language skills might also be ready factors
to explore the regression and threshold hypotheses.
Organized knowledge structure might have positively influenced the length of
acquired language skills maintenance compared with the non-organized. Inferred
from Chase and Simons Chase and Simon (1973) chunking
theory, the popular assumption is that organized knowledge improves memory by
consolidating more effective processing clues. For Chase and Simon, chunks were
integrated units in long-term memory involving elements sharing some dimension
of similarity. The learners richer experience in an area developed larger
chunks than those with less experience. Although important details might have
altered, the general mechanism of organizational processing continues to underlie
contemporary theories of skilled memory (Hunt and Rawson,
2011). According to long-term working memory theory (Ericsson
and Kintsch, 1995), individuals with a high level of knowledge are able
to use organized knowledge structures (referred to as retrieval structures)
to encode information quickly and reliably into long-term memory and subsequently
to access that information quickly and reliably from long-term memory (Hunt
and Rawson, 2011). Template theory (Gobet, 2000;
Gobet and Simon, 1996) also claims that long lasting
memory depends on the use of organized knowledge structures. According to template
theory, knowledge structures can include clusters (containing the location of
several chess pieces in a commonly occurring pattern) and templates (containing
information about fixed locations of around 10-12 pieces, with slots that may
be filled with other pieces or clusters in variable locations on the board)
to retrieval structures (to organize templates when more than one board configuration
must be remembered) (Hunt and Rawson, 2011). In this
study, some participants linguistic elements of negative sentences did
not suffer any attrition. They might have established some undetected organized
structure. For those who showed attrition in negation might have not successfully
formed this sort of structure. Therefore, when EF learners desire to keep the
acquired language longer, a profitable way might be to learn language in a more
systematic and organized way. Good learners tend to be those who have cultivated
a habit of learning language systematically in order to reach the threshold
so that attrition could be effectively prevented.
Findings in this study agreed to the threshold hypothesis since workers who
scored over 500 in CET4 performed significantly better than workers who scored
less than 400 and English teachers performed significantly better than most
workers and students. English teachers and workers who scored over 500 might
have reached the critical threshold and their reactivated level was lower so
that their knowledge in negative English sentences showed relatively steadier
and more easily reactivated. The regression hypothesis, however, was not fully
supported by this study. More empirical studies may be needed to test this anecdotally
convincing hypothesis. It seems necessary to conduct empirical studies before
we apply both regression and threshold hypotheses to the research of language
attrition. However, it is evidently painstaking to test the regression and threshold
hypotheses unless cross-disciplinary efforts are made. The validation may need
cooperation between many fields such as linguistics, psychology, neurology and
The author wishes to thank the people who help this study and the projects
which financially support this study: 2011 Youth Fund of Humanities and Social
Sciences of Ministry of Education of China The Regression and Threshold
Hypotheses of English Negative Sentences Attrition among English Learners in
China, (Project No.: 11YJC740138); 2011 Teaching Innovation Project of
Tongda College of Nanjing University of Posts and Telecommunications The
Regression and Threshold Hypotheses of Foreign languages and Teaching Innovation
of College English in Civil Colleges (Project No.: TD02011JG02); The Second
Batch of Post-doctoral Research Fund of Jiangsu Province in 2012 The Regression
and Threshold Hypotheses of English Language Attrition among Students in China
(Project No.: 1202112C)", 2013 Philosophy and Social Science Guidance Research
Project of Education Bureau of Jiangsu Province (Project No.: 2013SJD740005),
Special Fundamental Research Fund for the Central Universities (Project No.:
2013B33914), 2013 Shaoxing Important Research Project of Higher Education Reform,
and 2014 Research Project of Zhejiang Yuexiu University of Foreign Languages
(Project No.: N2014013).
Argaman, O. and S. Abu-Rabia, 2002. The influence of language anxiety on English reading and writing tasks among native Hebrew speakers. Lang. Cult. Curriculum, 15: 143-160.
Bailey, P., A.J. Onwuegbuzie and C.E. Daley, 2000. Correlates of anxiety at three stages of the foreign language learning process. J. Lang. Soc. Psychol., 19: 474-490.
Bardovi-Harlig, K. and D. Stringer, 2010. Variables in second language attrition: Advancing the state of the art. Stud. Second Lang. Acquisit., 32: 1-45.
Berko-Gleason, J., 1982. Insights from Child Language Acquisition for Second Language Loss. In: The Loss of Language Skills, Lambert, R.D. and B.F. Freed (Eds.). Newbury House, Rowley, USA., pp: 13-23.
Brown, R. and J. Kulik, 1977. Flashbulb memories. Cognition, 5: 73-99.
Cancino, H., E. Rosansky and Schumann, 1978. The Acquisition of English Negatives and Interrogatives by Native Spanish speakers. In: Second Language Acquisition, Hatch, E. (Ed.). Newbury House, Rowley USA.
Chase, W.G. and H.A. Simon, 1973. The Mind's Eye in Chess. In: Visual Information Processing, Chase, W.G. (Ed.). Academic Press, New York, USA., pp: 215-281.
Cheng, Y.S., 2004. A measure of second language writing anxiety: Scale development and preliminary validation. J. Second Lang. Writing, 13: 313-335.
CrossRef | Direct Link |
Cohen, A.D., 1989. Attrition in the productive lexicon of two portuguese third language speakers. Stud. Second Lang. Acquisit., 11: 135-149.
CrossRef | Direct Link |
Corder, S.P., 1967. The significance of learner's errors. Int. Rev. Applied Ling. Language Teach., 5: 161-170.
CrossRef | Direct Link |
De Bot, K. and B. Weltens, 1991. Recapitulation, Regression and Language Loss. In: First Language Attrition, Seliger, H.W. and R.M. Vago (Eds.). Cambridge University Press, Cambridge, UK., pp: 31-51.
De Bot, K., 2007. Dynamic Systems Theory, Lifespan Development and Language Attrition. In: Language Attrition: Theoretical Perspectives, Kopke, B. (Ed.). John Benjamins Publishing Company, Amsterdam, Netherlands, pp: 53-68.
Elkhafaifi, H., 2005. Listening comprehension and anxiety in the Arabic language classroom. Mod. Lang. J., 89: 206-220.
Ericsson, K.A. and W. Kintsch, 1995. Long-term working memory. Psychol. Rev., 102: 211-245.
Eysenck, M.W., 1979. Anxiety, learning and memory: A reconceptualization. J. Res. Personality, 13: 363-385.
Gobet, F. and H.A. Simon, 1996. Templates in chess memory: A mechanism for recalling several boards. Cognit. Psychol., 31: 1-40.
Gobet, F., 2000. Some shortcomings of long-term working memory. Br. J. Psychol., 91: 551-570.
Guo, J., 2010. Research about Chinese negation acquisition of South Korean children. M.A. Thesis, Normal University of Central China.
Hakansson, G., 1995. Syntax and morphology in language attrition: A study of five bilingual expatriate Swedes. Int. J. Applied Linguist., 5: 153-169.
CrossRef | Direct Link |
Hanson, L., 1999. Second Language Attrition in Japanese Contexts. Oxford University Press, Oxford, UK., ISBN-13: 9780195352641, Pages: 232.
Horwitz, E.K., M.B. Horwitz and J. Cope, 1986. Foreign language classroom anxiety. Mod. Lang. J., 70: 125-132.
CrossRef | Direct Link |
Hunt, R.R. and K.A. Rawson, 2011. Knowledge affords distinctive processing in memory. J. Memory Lang., 65: 390-405.
Jakobson, R., 1941. Kindersprache, Aphasie und Allgemeine Lautgesetze. Vol. 1, Almqvist & Wiksell, Uppsala, pp: 328-401.
Jordens, P., K. de Bot and H. Trapman, 1989. Linguistic aspects of regression in German case marking. Stud. Second Lang. Acquisit., 11: 179-204.
Jordens, P., K. de Bot, C. van Os and J. Schumans, 1986. Regression in German Case Marking. In: Language Attrition in Progress, Weltens, B., K. de Bot and T.J.M. van Els (Eds.). Foris, Dordrecht, Netherlands, pp: 159-176.
Keijzer, M., 2007. Last in First Out? An Investigation of the Regression Hypothesis in Dutch Emigrants in Anglophone Canada. LOT Publications, Utrecht, Netherlands.
Keijzer, M., 2009. The regression hypothesis as a framework for first language attrition. Bilingualism Lang. Cognition, 13: 9-18.
Kim, J.H., 2000. Foreign language listening anxiety: A study of Korean students learning English. Ph.D. Thesis, University of Texas, Austin.
Klima, E.S. and U. Bellugi, 1966. Syntactic Regularities in the Speech of Children. In: Psycholinguistics Papers: The Proceedings of the 1966 Edinburgh Conference, Lyons, J. and R.J. Wales (Eds.). Edinburgh University Press, Edinburgh, Scotland, pp: 183-219.
Kuhberg, H., 1992. Longitudinal L2-attrition versus L2-acquisition, in three Turkish children-empirical findings. Second Lang. Res., 2: 138-154.
Lambert, R.D., 1989. Language attrition. ITL Rev. Applied Linguist., 83: 1-18.
Larsen-Freeman, D. and M.H. Long, 1991. An Introduction to Second Language Acquisition Research. Longman, New York.
Liu, M., 2006. Anxiety in Chinese EFL students at different proficiency levels. System, 34: 301-316.
Livingston, R.B., 1967. Brain Circuitry Relating to Complex Behavior. In: The Neurosciences: A Study Program, Quarton, G.C., T. Melnechuk and F.O. Schmitt (Eds.). Rockefeller University Press, New York, USA., pp: 499-514.
MacIntyre, P.D. and R.C. Gardner, 1994. The effects of induced anxiety on three stages of cognitive processing in computerized vocabulary learning. Stud. Second Lang. Acquisit., 16: 1-17.
MacIntyre, P.D. and R.C. Gardner, 1994. The subtle effects of language anxiety on cognitive processing in the second language. Lang. Learn., 44: 283-305.
CrossRef | Direct Link |
Mills, N., F. Pajares and C. Herron, 2006. A reevaluation of the role of anxiety: Self-Efficacy, anxiety and their relation to reading and listening proficiency. Foreign Lang. Ann., 39: 276-295.
Milon, J.P., 1974. The development of negation in English by a second language learner. TESOL Q., 8: 137-143.
Direct Link |
Moorcroft, R. and R.C. Gardner, 1987. Linguistic factors in second-language loss. Lang. Learn., 37: 327-340.
CrossRef | Direct Link |
Neisser, U., 1984. Interpreting Harry Bahrick's discovery: What confers immunity against forgetting? J. Exp. Psychol. Gen., 113: 32-35.
Paradis, M., 2007. L1 Attrition Features Predicted by A Neurolinguistic Theory of Bilingualism. In: Language Attrition: Theoretical Perspectives, Kopke, B., M.S. Schmid, M. Keijzer and S. Dostert (Eds.). John Benjamins Publishing, Amsterdam, pp: 121-133.
Price, M.L., 1991. The Subjective Experience of Foreign Language Anxiety: Interviews with Highly Anxious Students. In: Language Anxiety: From Theory and Research to Classroom Implications, Horwitz, E.K. and D.J. Young (Eds.). Prentice Hall Inc., Englewood Cliffs, NJ., USA., pp: 101-108.
Schmid, M.S., 2002. First Language Attrition, Use and Maintenance: The Case of German Jews in Anglophone Countries. Benjamins, Amsterdam, Netherlands.
Slobin, D.I., 1977. Language Change in Childhood and History. In: Language Learning and Thought, Macnamara, J. (Ed.). Academic Press, New York, USA., pp: 185-214.
Song, Y., 2005. Effects of anxiety on listening performance and suggestions for improving listening teaching. CELEA J., 28: 11-17.
Direct Link |
Wode, H., 1981. Learning a Second Language. Gunter Narr, Tubingen, Germany.