plik

YEAR TWO

Introduction to SLA

#6: The Significance of Learners' Errors

1. The theory of transfer - The paradigms of Charles Osgood

Osgood (1949) summarised two decades of research into the phenomenon of transfer in the three
'paradigms':

Paradigm

Task 1

Task 2

Task 3

Transfer-value

-T

RETROACTION

PROACTION

Osgood  envisaged  three  learning  tasks  being  set  in  sequence:  notice  that  for  each  paradigm  (A,  B,  C)
task 1 and task 3 are identical. When considering the effects on task 2 of having already done task 1, we
speak  of  Proaction,  whereas  Retroaction  is  concerned  with  the  effect  of  having  done  task  2  on  a
previously  learned  activity.  In  fact,  there  are  only

two learning tasks, not three: task 3 is in reality a

performance task. CAH is concerned with proaction of course, seeing task 1 as the learning of L1 and
task 2 as the learning of L2. Retroaction is of potential interest to CAH in two ways: first, it could handle
effects of L2 upon performance in L1, known as backlash. Secondly, it is concerned with forgetting. It
would have to be invoked in any attempt to explain why L1 is not usually forgotten when a L2 is learned.
Here, we shall only be concerned with proaction.

Osgood assigns Transfer values to each paradigm: +T is positive transfer or 'facilitation' while -T is
negative transfer or 'interference'. The amount of +T or -T generated by each paradigm will depend, of
course, on how similar the stimuli (Ss) are with identity of the responses (Rs) or how similar Rs are with
identity of Ss.

The native language of learners exerts a strong influence on the acquisition of the target language
system.

While the native system will exercise both facilitating and interfering effects on the

production  and  comprehension  of  the  new  language,  the  interfering  effects  are  likely  to  be
most  salient.  […]  Errors  are  windows  to  a  learner's  internalized  understanding  of  [L2],  and
therefore  they  give  teachers  something  observable  to  react  to.

Student non-errors - the

facilitating effect - certainly do not need to be treated. Don't try to fix something that isn't
broken.

(Brown 2000: 66;

emphasis added)

2. Criticisms of the Contrastive Analysis Hypothesis and the emergence of Error Analysis

 empirical: doubts concerning the ability of CAH to predict errors arose when researchers began to

examine language-learner language in depth

 theoretical: there were a number of theoretical reservations concerning the feasibility of comparing

languages and the methodology of CAH

 practical: there were doubts about whether CAH had anything relevant to offer to language teaching

Gradually, in the heyday of the audiolingual approach to foreign language teaching in the 1960s, which
advocated avoiding errors at all cost, researchers (e.g. Corder 1967) began to realise not only that
mistakes were a natural and inevitable outcome of learning, but that a

systematic study of learners'

errors could help us discover the processes underlying second-language acquisition. The new approach
paved the way for Error Analysis (EA for short).

The procedure for EA, spelled out by Corder (1974a), is as follows:

i. a corpus of language-learner language is selected (with regard for the size, the written vs.

spoken medium, the learners' age, L1 background and level of proficiency)

ii. the errors in the corpus are identified (e.g. performance slips vs. errors of competence)
iii. the errors are classified (e.g. according to a linguistic typology)
iv. the errors are explained (e.g. the psycholinguistic causes of errors are given)

v. the errors are evaluated (e.g. by assigning the seriousness of each error in order to take the

principal teaching decision, but only if a given EA is pedagogic; if it serves L2 acquisition
research, this stage is redundant)

From an EA perspective, the foreign-language learner is no longer perceived as a passive recipient of
input, but rather as playing an active role, processing input, generating and testing hypotheses.
According to EA, learners' errors are crucial because they indicate progress in developing the system of
L2.

3. Empirical research and the predictability of errors

The criticisms levelled against CAH in the late 1960s and early 1970s triggered off more interest in L2
learner errors. Some of the conclusions drawn from the findings of empirical research were so extreme
that they invalidated the predictions made earlier by contrastive studies. For example, Dulay & Burt
(1973; 1974) identified types of error according to their psycholinguistic origins:

 85%: first-language developmental errors, i.e. those that do not reflect L1 structure but are found in

first-language acquisition data

 12%: unique errors, i.e. those that do not reflect L1 structure and also are not found in first-

language acquisition data

 3%: interference-like errors, i.e. those that reflect L1 structure and are not found in first-language

acquisition data

However, other research did not bear out Dulay & Burt's findings: in the studies relevant for EA and
published in the 1970s and early 1980s, the percentage of interference (i.e. negative transfer) errors
varies from 30 to 50%.

4. Error typology and error sources

An  error  can  be  defined  in  a  number  of  ways.  For  example,  George  (1972)  describes  it  as  a  form
unwanted by the teacher; a more recent definition, on the other hand, describes it as "a deviation from
the  norms  of  the  target  language"  (Ellis,  1994:  51).  Early  attempts  at  error  typology,  such  as  those
based  on  CAH  and  behaviourism,  were  rather  simplified  and  distinguished  roughly  between  negative
transfer  (interference)  and  positive  transfer  (no  error,  despite  the  effect  of  the  old  habit  on  new
learning).

A  distinction  is  often  made  between  errors  "of  competence"  and  mistakes  "of  performance",  whereby
the  former  are  instances  of  genuine  gaps  in  linguistic  knowledge  or  noticeable  and  consistent
dissentions from the grammar of an adult native speaker, while the latter are mere random guesses or
slips,  i.e.  failures  "to  utilize  a  known  system  correctly"  (Brown,  1994:  205).  A  third  option  has  been
suggested  by  Edge  (1989),  who  came  up  with  the  notion  of  attempts  -  efforts  by  learners  who  are
experimenting with the interlanguage system and predicting the rules and forms that they have not yet
studied formally.

Interlingual sources of errors refer to differences between L1 and L2 (Johnson, 2001: 66). In simpler
terms, these errors occur as a result of using some elements from one language while speaking another.
They can be further subdivied into:

 overextension of analogy, in which the learner misues some item because it shares features with an

item in L1 (e.g. English *

process to mean ‘trial’, as in Polish proces);

 transfer of structure, which arises when the learner utilises some L1 feature (phonological,

grammatical, lexical or pragmatic) rather than that of the target language. This is what is generally
understood by transfer;

Intralingual errors fall into the following categories (Johnson, 2001: 66):

 overgeneralisation errors, which occur when the learner creates an incorrect structure on the basis

of other structures in the target language. It often involves one deviant structure in place of two L2
structures, e.g. *

He can sings coming from He sings and He can sing;

 ignorance of rule restriction, which is closely related to the previous category and involves the

application of rules to contexts where they cannot apply, e.g. *

He made me to do it;

 incomplete application of rules, which involves a failure to fully develop a structure. Thus, learners

of English as a L2 can use the wrong word order, e.g. in questions such as

You want to dance?, where

intonation questions (i.e. declarative statements with rising intonation) are used instead of

yes-no

questions;