Error Analysis (EA)

Error Analysis (EA) is a type of linguistic analysis that focuses on the errors learners make. Unlike Contrastive Analysis, L1 is not used to compare. Comparison is made between learners' errors in TL and the TL form itself. It is similar to the weak version of contrastive analysis in that both start from learner production data; however, in contrastive analysis the comparison is made with the native language, whereas in error analysis it is made with the TL.

Corder (The significance of learners’ errors, 1967) observed that learners' errors may be significant in themselves, and they should not be discarded as such. This led to shift emphasis from pedagogical issues to others. Errors can be indicators of learner's knowledge of L2. It has been found that errors are not a reflection of faulty imitation. But errors are indicators of learners' attempts to bring a system into the new language being learned. Probably, focus on errors led to the beginnings of the field of SLA. It is important because of its implications on psychology and linguistics also.

Corder identifies the difference between errors and mistakes. Mistakes are like the slips of the tongue- one time only events. Speaker corrects it after identifying it. An error is systematic. It occurs repeatedly, ad is incorporated in the language system of the learner. Therefore, these are errors according only to the teacher, not according to the learner. For the learner, it is all part of the IL. For example, utterances like 'no speak', 'no understand', etc. are consistent and systematic errors of the learner, and are mistakes according to the teacher! Such interpretations could hold learning/teaching process back.

Error analysis was done inside classrooms with pedagogical remediation as goal, using the following steps:

1. Collect data: written and oral
2. Identify errors
3. Classify errors
4. Quantify errors
5. Analyze source
6. Remediate: Based on the kind and frequency of an error type, pedagogical intervention is carried out.

Error analysis is more useful to teacher/researcher to explain errors. There are 2 types of errors, interlingual and intralingual in EA. Interlingual errors can be attributed to the NL - cross-linguistic comparisons. Intralingual errors stem out from the TL, independent of the NL. One would expect similar intralingual errors from speakers of varieties of languages. 

Criticism

1. One criticism said that EA is all about errors, and that one should consider errors as well as non-errors to get the entire picture of a learner's linguistic behaviour. 
2. A 1974 article by Schachter showed that the NL is a determining factor in accounting for the facts of restrictive relative clause production, yet these facts would not be apparent through an error analysis alone! She studied the use of Restrictive relative clauses in English by native speakers of Persian, Arabic, Chinese, and Japanese. The error data she collected would say that Chinese and Japanese learners of English had control over formation of restrictive relative clauses, and Persian and Arabic users do not. But further data on errors plus non-errors showed a different analysis. While error were more in Persian and Arabic learners had more errors, they also produced half as many correct restrictive relative clauses as Japanese and Chinese learners! Why does this discrepancy occur?
   Japanese and Chinese form relative clause by placing the modifier before the noun it modifies. Persian and Arabic relative clauses are similar to English in that the relative clause is placed after the noun it modifies. Because the difference between how NL and TL forms relative clauses, learners do not frequently use the structure (Chinese and Japanese). But when they use it, they use it cautiously and with high degree of accuracy. Persian and Arabic learners use them a lot because their NL structure is similar to TL structure of relative clauses, and therefore, make more errors. Thus, EA alone couldn't bring out the explanation just by looking at the errors, while in fact, NL was a major factor.
3. Another difficulty is in judging if something is an error. Learners can use structures of NL to construct sentences in TL. It might be interpreted as an error of some kind, while it might be of another kind. So, there could be a mismatch between what the teacher judges as error and what the learner is actually attempting to do. 
4. Attempt to give reasons to errors is another inadequacy or EA. The assumption is that is the form is correct, underlying rule is also correct. But learner might make correct sentences, yet may not have internalized the necessary background rules.  In sum, error analysis alone cannot provide us with this information, because an assumption of error analysis is that correct usage is equivalent to correct rule formation.
5. Source of errors also pose criticism. EA says that errors can be categorized as belonging to one source or another. Can we attribute single reason for errors? Learner production may be influenced simultaneously by multiple sources (article system and Czech learners of English). Source of error could be TL and NL simultaneously also. Schumann studied the use of negation by Spanish learners of English. Learners pass through 5 stages before figuring out that 'do' is the element that carries tense and person distinction qualities in negation. He observed that certain stages of development are more persistent for learners from certain languages. He found that in case of Spanish learner of English, 2 forces, namely native language and facts of development act as sources of error. In case of learners from other languages, the only factor at play is development. He says that a single source of error will have less influence than a set of converging sources, and will lead the learner to move much more rapidly in the developmental sequence. 

EA acknowledges that learners are more than imitators of language. But it only sees a part of what a second language learner produces. Therefore, it doesn't analyse sufficient data. EA doesn't have comprehensive approach. Therefore, one cannot hope to understand a learning situation with a partial study like EA.

Notes prepared from: Gass, Susan M. (2008). Second Language Acquisition. Routledge, New York.

Contrastive Analysis Hypothesis

1950s and 60s saw language as a habit. Second language learning was seen as forming a new set of habits. Therefore, native language had a very relevant role, since in this view of language learning, it was the major cause of lack of success in language learning. The habits established in childhood (NL) interfered with the establishment of a new set of habits (TL). From this understanding emerged the need to compare NL and TL. This is known as Contrastive Analysis which compares the rules of two languages to determine similarities and differences. Robert Lado is the major proponent of the this field.

Contrastive Analysis (CA) is a way of comparing languages in order to determine potential errors to isolate what needs to be learned and what doesn't need not be learned in second language learning. Phonology, morphology, syntax, social aspects, etc. are studied to predict what will be easier and difficult for learners. Similar structures will be easily transferred and learned.

Lado says:

Since even languages as closely related as German and English differ significantly in the form, meaning, and distribution of their grammatical structures, and since the learner tends to transfer the habits of his native language structure to the foreign language, we have here the major source of difficulty or ease in learning the structure of a foreign language. Those structures that are similar will be easy to learn because they will be transferred and may function satisfactorily in the foreign language. Those structures that are different will be difficult because when transferred they will not function satisfactorily in the foreign language and will therefore have to be changed. (Lado, 59) 

Pedagogical materials that came out of CA in North American tradition were based on the following assumptions:

1. CA is based on the assumption that language is a habit, learning a new language is establisment of a new set of habits. 
2. major source of error in second language is the native language - NL.
3. errors can be explained using differences between NL and TL.
4. Greater the difference, greater the difficulty. 
5. One has to learn the differences. Similarities are easily transferred.
6. difficulty and ease are proportional to differences and similarities between the two languages under consideration. 

Various views on CA Hypothesis

Strong/a priori/predictive view: One can predict about learning, and success of materials based on CA. 

Weak/a posteriori/explanatory view: Starts with the learners' recurring errors and gives explanations for the learner behaviour based on CA.

Weak version gained faith because the strong version failed. Weak version gave importance to the learner, the forms they produced and the strategies they used to reach their IL forms.

CA did not survive because its theoretical background-behaviourims- belief that NL was the driving force of L2 learning- was discarded. In the 60s, language came to be seen in terms of structured rules. Behaviourism was discarded. Learning was no more seen as imitation and habit formation, but as active rule formation. 

The failure of behaviourism had implication on SLA. If imitation and reinforcement has no bearing on NL acquisition, may be SL also is not influenced by it. This became evident through data analysis. Some errors learners produced in L2 were in no way related to the structures/errors in their L1. (He comed yesterday- attempt to impose regularity on irregular verb). The theory did not predict what was happening in non-native speech. Not only did the predictions NOT come true, things that they did not predict appeared more than often. Within a theory based on the transference of NL forms, this could not be explained, for why should transfer occur in one instance, but not in another?

For example:

In French, object pronouns precede the verb, as in 

- Je  les  vois.

  I  them see

 "I see them.”

In English, object pronouns follow the verb. However, the following facts emerge in learner data:

By French learners of English 

I see them. (produced) 

*I them see. (not produced) 

By English learners of French- None of these is possible in French. 

a. Je vois elle. I see her. 

b. Le chien a mangé les. The dog has eaten them. 

c. Il veut les encore. He wants them still. 

In other words, French learners of English never prepose the object pronoun. Rather, they correctly follow English word order, which in this case is in violation of French word order. With English speakers, the reverse occurs: they follow the native language word order. If the “habits” of one’s native language are the driving force, then why should they be operative in one language, but not the other? (Gass 98-99)

The ideas of difficulty were also questioned. Difficulty was equated to errors in CA. Error meant that learner was having difficulty in learning. It is not a real measure of difficulty. How does one judge what is difficult for the learner? Error is not a real measure of difficulty. To equate difference with difficulty attributes a psycholinguistic explanation to a linguistic description. 

We can't say that there are no factors in NL that influences TL. But there surely are other factors than NL. The conclusion is that the 1:1 correspondence implied by CA Hypothesis between native and second language does not hold ground. It is not that simple. L1 has its effect, but cannot be limited to difficulty and transfer. There are other factors that may influence the process of acquisition, such as innate principles of language, attitude, motivation, aptitude, age, other languages known, and so forth. 

Comparing languages is a complex business. Lado himself had identified it. Stockwell, Bowen, and Martin gives a framework or hierarchy of difficulty in learning. it speaks also about ways in which languages can differ. Categories in which there is differentiation (NL has one form, TL has two forms), absence of some category in either languages (articles in English; Japanese has no articles), Only one form in L2, but two in L1, Correspondence, etc. are the elements of the hierarchy. CA also failed to validate claims through data from real world (empirical basis). 

Lado's hypothesis inspired a lot of research in the field of second language learning (to match CA predictions and actual data). As a result of Lado's warning to check hypothesis against actual data, Error Analysis emerged.

Reference
Gass, Susan M. (2008). Second Language Acquisition. Routledge, New York.
Secondary: Lado, R. (1957). Linguistics Across Cultures. Ann Arbor: University of Michigan Press.

Psychological background of language acquisition, Contrastive Analysis

Language learning and Associate theories have derived their Technology from Psychology of learning. Behaviourism was the leading school of thought at that time. One of the key concepts of behaviourism was the notion of transfer. Concept says that if you learn concept A first, then all the concepts that you learn after that concept will be influenced by the first concept. If someone has learned tennis, she will use this knowledge when she learns to play table tennis. Old knowledge is thus transferred to new situation.

Learning is cumulative for behaviourists. The more knowledge someone has, the more likely it is that her learning is influenced by her past experiences and learning. An adult rarely finds something 'completely' new. 

The implication is that speed of learning can be influenced by what you already know. That is old knowledge will be transferred to new situation. Behaviourist notion underlines Expectations of habit and cumulative learning. For an adult the point of departure is learning of his/her past.

Let us now see how this concept is applied to second language learning. The concept of transfer has two aspects. Positive transfer or facilitation and negative transfer or Interference- these are not two distinct cognitive processes, but points to whether the process gives correct or incorrect results.

Example: If a Spanish speaker is learning Italian, when asking a question that speaker might correctly produce - Mangia bene  il   bambino?      eats     well the   babybecause in Spanish one uses the same word order to form questions.- ¿Come bien el   niño?       eats  well the babyThis is known as positive transfer. But if that same speaker is learning English and produces- Eats well the baby?the incorrect utterance is known as negative transfer. (Gass, 113)

There are two types of interference. Retroactive inhibition and proactive inhibition. Retroactive inhibition is when someone forgets something that was learnt earlier because of the new learning - language loss. Proactive inhibition is when previously learned responses appear in situations where new ones are required. This is more similar to second language learning since the first language in this model influences the learning of the second language.

But most of the experiments and results are gathered inside the laboratory. Their application in the real world has to be meticulously tested. 

Lado was the one who brought the theoretical underpinnings of behaviourist position explicit. Now turning to the work on second language acquisition based on the behaviourist positions, the major reason behind all this work was pedagogical. Behaviourists believed that problems in second language acquisition arise not out of difficulty in the features of language, but out of the special set created by the first language  itself.

Thus in the 1950s and 60s, language was understood as habit. Second language learning is the development of a new set of habits. And native language was seen as an impediment to learning L2. Childhood habits interfered with the establishment of new set of habits. 

From this framework, contrastive analysis (CA) emerged, because one needs to compare the rules of two languages. There are two traditions in contrastive analysis- North American and European. North American tradition emphasized on language teaching and learning. The goal of analysis was improvement of class room materials (Applied contrastive analysis according to Fisiak). In European tradition, goal was not pedagogical, but to gain a greater understanding of languages. Here, CA is a sub-discipline of linguistics.

Reference

Gass, Susan M. Second Language Acquisition, Routledge, New York 2008.

Behaviourism in (First) Language Learning

The role of native language has come to be known as language transfer. Much of the theory in this field is connected to SLA research. Language transfer is accepted or rejected because the associated theory is accepted/rejected. The assumption always was that the second language learner relies heavily on her/his native language. The need to produce pedagogically relevant materials prompted scholars to make contrastive analysis of native and target languages to determine their similarities and differences.

Transfer in this context should be explained. This can be determined based on output. That is, although the use of the term implies a process, the result is determined by the product.

Behaviourism 
Bloomsfield's classic work 'Language' (1993) provides the most elaborate description of the behaviourist position with regard to language.

Typical behaviourist position is that language is speech primarily. It is a precondition for writing. It is believed so because children learn to speak before writing, and some societies have no written language but all have spoken language, there are no societies with only written language.

Speaking is imitation and analogizing. As children we establish habits, and grow them by analogizing from what we already know or mimicking the speech of others.

Bloomfield's description of how language acquisition takes place:

1. babbling generated by a child- imperfect repetition of something according to Bloomsfield. Sounds are imitated, resulting in habit formation. This babbling trains it to reproduce vocal sounds which strike its ear.
2. Next is pairing this stimulus with the response of a native speaker. Mother saying something to the child initiates the response based on its habit through babble.
3. Mother says doll when handing the child a doll. Thus handing of the doll, hearing of 'doll' and sight of the doll happens until it becomes a habit- sight and feel of doll makes it say 'doll'.This is how language is learned.
4. Bloomsfield says that the absence of the stimulus can generate another stimulus, which in turn can generate the desired/same response.
5. Correct performance yields better results. If child utters something very vaguely, adults won't understand.

In short, child learns to make stimulus-response connection. Learning involves the establishment of a habit by means of which these stimulus-response sets become associated.