Error analysis

Definition
Error analysis is a branch of applied linguistics. It is concerned with the compilation, study and analysis of errors made by second language learners and aims at investigating aspects of second language acquisition.

Closely related to error analysis is the concept of interlanguage.

Some researchers distinguish error analysis from transfer analysis, which compares the learner’s data with the respective first language, whereas error analysis compares the learner’s data with the target language norm and identifies and explains errors accordingly (cf. James 1998).

Development
Error analysis was first used as a way of studying second language acquisition in the 1960s. Corder’s seminal paper "The Significance of Learner’s Errors" (1967) had shifted researchers’ attention from the teaching perspective to the learning perspective – and therefore also away from contrastive analysis, behaviorism and structuralism towards cognitive psychology. This development went hand in hand with the turn towards a communicative approach in language teaching.

Drawing on knowledge about first language acquisition, Corder posited that second language learners discover the target language by hypothesizing about it and testing their hypotheses more or less like children do. This process does not happen randomly, but follows the learner’s built-in syllabus, so that errors will necessarily be made.

Corder used the term transitional competence for what has since become a widely accepted and often used concept: that of interlanguage (cf. Selinker 1972), the learner’s individual, dynamic approximation of the target language. According to this view, errors indicate that a learner actively learns the target language, as they occur whenever a hypothesis tested by the learner does not work. In error analysis, the language learning process is regarded as being influenced by the learner’s first language, his or her interlanguage and the target language. Thus, all of these three language systems have an influence on which errors a learner makes. But the gap between the interlanguage and the target language is considered the most important factor of the three. Even more importantly, however, the learner makes errors because of the learning strategies he or she employs to ‘discover’ the target language.

For all these reasons, inductive error analyses were carried out in order to arrive at generalizations about errors, interlanguage and, ultimately, second language acquisition. Error analysis reached its zenith in the 1970s, but soon turned out to be deficient as a research tool. By the late 1970s, it was merely contributing to broader second language acquisition theory and research, as it still does today.

Aims
The primary aims of error analyses were (i) to identify types and patterns of errors and (ii) to establish error taxonomies. These were supposed to be used to describe interlanguage and its development, i.e. the learner’s internal syllabus. Common difficulties in second language acquisition were to be identified. On this basis, error analysis was supposed to contribute to a comprehensive knowledge about processes of second language acquisition -- always assuming with Chomsky that there is something like a language acquisition device.

In addition, results were intended to be used for a revision of theories of language learning as well as help to evaluate and improve language teaching.

Results
The main achievement of error analysis consists in a change of perspective. Firstly, it let learners’ errors appear in a new light. They were no longer regarded as "signs of inhibition" (Corder 1967) that needed to be eradicated. Instead, they were regarded as useful “evidence of [...] strategies of learning” (Corder 1967) and as perfectly natural aspects of second language acquisitin. Secondly, it widened the perspective on possible causes of errors. Researchers recognized that the first language is not the only – in fact, not even the most important - factor that can lead to errors.

Common errors typical of different target languages were identified and, in search of reasons why those errors were made, they were classified in a new way. Errors were distinguished from mistakes or lapses, which are performance errors that are not determined by the interlanguage but rather by situational factors such as tiredness. Only ‘true’ errors are connected to the state of the interlanguage, or the learner’s competence. Interlingual errors, a result of interference from the native language, were differentiated from intralingual errors, occuring for example when a target language rule is applied to areas where it is not applicable. Corder also pointed out that an utterance which is seemingly correct but does not mean what the speaker or writer intended it to mean contains, in fact, a covert error.

Error analysis also played an important role in the development of the interlanguage hypothesis.

Criticism
Error analysis has been criticized for a number of practical problems, all of them connected to the fact that it tries to gather knowledge of language learning processes by examining the learner’s output. First of all, it has proved difficult to determine whether there is an error at all, and if so, what exactly constitutes it. The distinction between error and mistake cannot easily be made either. Secondly, there is usually more than just one way to classify an error. Thirdly, causes of errors are difficult to identify; there is a multitude of possible causes (e.g. communication strategies, personal factors, external factors), and since the learner’s output is the only source of evidence used, found causes are necessarily unreliable. In addition, “error taxonomies often confuse description with explanation” (Johnson & Johnson 1998:112), thus providing little to help learners.

Other criticism has aimed at the simplistic approach that error analysis takes toward second language acquisition. Only looking at incorrect output and ignoring correct output as well as any other aspects of the learning process means leaving out important sources of information that could be used to describe the acquisition process. This is related to the fact that correct output does not necessarily imply that something has been learned – among other reasons, because the learner’s language production varies in several ways.

As a result, error analysis has been subject to criticism. For example, it has been claimed that what was called ‘universal’ errors (errors that are made by any learner of a given target language, no matter what the first language) might in fact be interference errors (Byram 2004, cited in James 1998).