Analysing and counteracting counteracting interference interference errors David Lott There are are obvious advantages for teachers in conduc ting their o wn error analysis research: they can find ou t why their stud ents are making errors and then plan appropriate remedial lessons. But teachers wanting to undertake their own research are likely to have problems with the confusing glut of information on error analysis, and the lack of information on how to apply the results. Aft er d is cu ss ing so me of th e areas o f co nt rad ic tio n and co nt rov ersy in th e fi eld o f err or analy si s, th e arti cle g oes o n to des cr ibe an interference error analysis proj ect, including a detailed definiti on of interference error. Finally several several approaches to ‘teaching ou t’ interference errors are outlined.
In the course of teaching English to Italian students, I began to think about the extent to which interference (or ‘transfer’) from their native language was disrupting the students’ attempts to master English. A colleague and I conducted some research into the errors that our students made, in the light of current theories about second language acquisition. acquisition. Finally we experimented experimented with various ways of ‘teaching out’ interference errors. Below I shall seek to give a clear definition of interference, by identifying varieties of interference errors, but in the interim, I will define interference errors as errors in the learner’s use of the foreign language that can be traced back to the mother tongue.
The value of error analysis It is only comparatively recently that researchers have begun to develop hypotheses hypotheses about the process of second language acquisition. acquisition. This work has had obvious advantages for teachers, in particular in offering them techniques which they can use to analyse their students’ errors. Teachers can begin to attribute a cause to an error with some degree of precision. They can also build up a picture of the frequency of types of error; thus they can find out whether, for example, mother tongue interference, or teaching techniques, or problems inherent in the target language are the major cause of their students’ errors. In this way it is possible to plan classes giving very specific help to the students. However, a teacher wanting to carry out such a procedure has serious hurdles to overcome: there is a glut of information on techniques of error analysis, and a dearth of information on ways of applying the results. This makes both analysing errors and teaching on the basis of the results a difficult business.
Controversy The conflict of views and the contradictory results in the field of second language acquisition acquisition and error analysis can be attributed to the complexity of errors, to our limited knowledge of the psychological and the neurological processes involved in language learning, and to the wide scope for variety in a number of parameters: type type of teaching, age, attitude, attitude, and motivation motivation of the students, students, method method of elicitation, elicitation, and degree degree of relatedness of the native language and the target language. The variety of error categories that have been proposed presents the intending intending researcher with immediate immediate difficulties. Corder (1973:271) suggested that there are three basic categories of error: 1 pre-systematic errors: i.e. those made by a learner while he or she is trying to come to grips with a new point; 2 systematic errors: i.e. those which occur when the learner has formed an inaccurate hypothesis hypothesis about the target language; 3 post-systematic errors: i.e. the temporary forgetting of a point that had been previously previously understood. Selinker (1972:216-21) proposed nine different types of error, mainly systematic. These include: a. language transfer (items and rules in the learner’s version of the target language which can be directly traced back to the native language) language) ; b. transfer of training (the error is directly traceable to some fault in the teaching). Linguists Linguists have drawn a distinction between ‘competence errors’ and ‘performance errors’, which might be likened to Corder’s ‘systematic’ and ‘post-systematic’ ‘post-systematic’ errors respectively. respectively. This has led to a distinction distinction between ‘errors’ and ‘mistakes’. Also in the wake of transformational transformational generative grammar, linguists have drawn a distinction between ‘deep structure’ and ‘surface structure’ errors. As many have remarked, the existing categories of error are often difficult to use when attempting attempting to analyse students’ actual errors. Part of the problem lies in the difficulty difficulty of judging judging the cause of error with with some precision. precision. Thus, Thus, when attempting attempting to use an elaborate classification system such as Selinker’s, the problems are compounded and the researcher needs to have an extremely good insight into the learner’s mind. McDonough remarks that it should not be supposed that all learners take the same route to the same error. ‘Nor should it be assumed that one learner may not, at different times, produce the same error for different reasons’ (1981:115). There are further difficulties in categorizing errors that arise from conflicting interpretations interpretations of key concepts. For example, learners’ errors in the target language may be due to the non-existence of an item or rule in their native language. For example, Italian students have difficulty with three of the ways of referring to the future in English (e.g. I’m going to stay, I will stay, Z am staying), because such a distinction of form and usage does not exist in Italian. Can this be defined as an interference error? It might be argued that Italian gives no guide in this case but does not actually ‘interfere’ with the learning of English, and that the error is simply a confusion of the rules of the target language. There is considerable controversy over the definition of interference. Dulay and Burt (1976:71) define interference as ‘the automatic transfer, due to habit, of the surface structure of the first language onto the surface of the target language’. They would presumably not consider the above example as an interference error. Nickel takes a wider view of interference and would consider that the above error is due to interference from the native language, though he suggests in such cases that: ‘One will have to distinguish . . . between direct and indirect interlingual interferences. It may be that what one can call the “macro-cause” is interlingual and what one might call the “micro-cause” is intralingual’ (1981:9). Error analysis projects have yielded wildly contradictory results. No doubt this is in part due to differences of definition. At one extreme, Dulay and Burt stated that out of 513 errors made by Spanish children learning English, English, fewer than 5 per cent could be classified as interference errors. At the other extreme, Nickel observes that some comparative linguistics linguistics researchers may have attributed as many as 80 per cent of errors to
interference, while ‘the more realistic ones’ would have voted for an average of about 60 per cent. In our own research we found that approximately 50 per cent of errors were due to interference. To sum up, a teacher wanting to analyse students’ errors is likely to have problems with : 1 the considerable variety of error types that have been proposed; 2 the controversial nature of key concepts (e.g. ‘interference’) ; 3 the contradictory results of published error analysis projects. It would seem helpful, given the general disagreement among researchers, to suggest guidelines that teachers might follow in conducting their own error analysis.
Analysing interference errors As it is not uncommon for researchers to attribute about 50-60 per cent of errors to interference, we recommend that teachers, at least initially, concentrate on getting information on interference errors. Hypothetically, most teachers should thus be in a position to focus on at least half of their students’ errors. There is also an obvious advantage in keeping the category choices to a minimum, given the problems in selecting the correct error categories. Needless to say, if it becomes apparent that in the course of the analysis not many errors can be attributed to interference, then an alternative category or categories should be considered. The choice of material for analysis should be carefully considered. In our research we analysed written English in a selection of examination papers. Nickel suggests that free conversation seems to produce different frequencies and qualities of error from those in written tests (1981:12). However this is not borne out by our experience. He also asserts that there seems to be general agreement that there is more transfer in translations than in other types of test; while multiple choice and cloze tests produce fewer contrastive errors (1981:12). Thus, suitable corpora would be reasonably free compositions or spontaneous conversations. In a systematic analysis of the written English of a large number of students, one might select the work of ‘average’ students for analysis. In our research we selected the examination papers offered by students who just achieved the pass-mark and no more. To be able to categorize errors without great difficulty it is essential to have precise definitions. In our research we considered an error as due to interference from the native language if it fulfilled one of the following criteria : 1 Overextension of analogy: The student misuses a vocabulary item because the item shares features, whether phonological, orthographic, semantic, or syntactic, with an item in the native language. For example: a. Eng. ‘process’ is used to mean ‘trial’, because of the phonetic and orthographic similarities with It. ‘processo’, which does mean ‘trial’. b. Eng. ‘work’ is used to mean ‘job’, as in ‘I have a good work’, because of the wider semantic reference of It. ‘lavoro’, which can be used for both work in general and a job in particular. c. Eng. ‘mad’ is used as a noun, as in ‘Stop driving like a mad !‘, because of the greater grammatical flexibility of It. ‘pazzo’, which occurs as both an adjective and a noun (madman). 2 Transfer of structure: The student makes an error of grammar because he or she is following the rules of the native language and not the rules of the target language. One student in our sample wrote: ‘My parents arrived, we decided to go home’. This reflected the different rules governing the use of the past participle in English and Italian.
3 Interlingual/intralingual error: a. The student makes an error of grammar because a grammatical distinction does not exist in the native language. In our sample we found students had difficulty with the present perfect (I have gone) as distinct from the simple past (I went), because in colloquial Italian the equivalent forms are interchangeable in most contexts. b. The student misuses a vocabulary item because a lexical distinction does not exist in the native language. For instance, a student wrote: ‘The company makes business in India’; the ‘make/do’ distinction does not exist in Italian, which would translate both ‘make’ and ‘do’ as ‘fare’. It is obviously impossible to build a system of definitions where the analyst will not periodically have doubts about how to categorize particular errors ; the quality of any research must be affected by the researcher’s intuitions about why the error occurred. The inclusion of the third category might be considered to indicate too broad a view of interference, but by taking a fairly broad view, the teacher can get a complete picture of how the mother tongue is affecting the learning of the target language.
Procedure 1 We systematically categorized the errors into interference and noninterference errors, keeping a count of both types so that at the end of the analysis we could calculate the proportion of errors caused by interference. 2 We made a note of the individual interference errors. 3 We classified the interference errors into the categories outlined above - i.e. overextension of analogy, transfer of structure, interlingual/intralingual errors. (The reason for this will become apparent when we consider ways of applying the results in the classroom.) 4 At the end of the analysis we checked which grammatical errors were repeated. In our analysis we found most of the errors conformed to a surprisingly small number of types: e.g. tenses, prepositions, collective nouns, and articles. 5 We attempted to draw a distinction between ‘performance’ and ‘competence’ errors; we considered only those error types that occurred more than three times in the sample as worthy of inclusion in the teaching programme.
Applying the results If the results of the research indicate that a significant proportion of errors can be attributed to interference, considerable advantage should accrue from attempting to ‘teach out’ interference errors. But how? McDonough suggests that the analysis of errors can be the basis for a ‘guided discovery’ technique of teaching (1981:124). In other words, the teacher asks the student to complete a series of sentences incorporating a particular problem, the exercises being organized so that potential mistakes are provoked and guidance is given by the teacher. In this way the teacher helps the student to make valid hypotheses. A colleague and I experimented with a variety of methods for eliminating grammar and vocabulary errors, including ‘guided discovery’. This proved to be effective in a number of ca ses, but certainly not in all. It quickly became apparent that the variety of error types required a variety of remedial strategies. We found that the categorization of errors as transfer of structure errors, interlingual/ intralingual errors, or overextension of analogy errors was a useful starting point.
Interlingual/ intralingual Since these errors are caused by the lack of a distinction in the native language, there was no call to make any explicit comparison with the native
language. The problem was to establish lexical and grammatical distinctions in the learners’ conception of English. In many cases the most effective method was the ‘guided discovery’ technique. Thus, in teaching the difference between ‘make’ and ‘do’, we asked the students to consider two sets of sentences : Set 1: I’m sorry, I’ve made an error. They always make a lot of noise. etc Set 2 : He likes doing the cooking. She does the cleaning at the weekend. etc. The students had to decide what the basis for the difference between Sets 1 and 2 was, and try to apply it to a series of mixed sentences. As the students worked through the sentences, we helped them to develop and improve their hypotheses. In a general consideration of language teaching techniques, McDonough suggests that where the rule is complicated, an imposed rule may have advantages over a rule discovered by the student (McDonough 1981:34). Thus to help our students use the simple past and the present perfect in English correctly, we began with an explanation of the grammar, using diagrams or pictures, and the contrasting usage was exemplified in a dialogue. We then asked the students a series of questions involving the language point: How long have you done this job? What did you do before that ? Describe yesterday’s lesson. Describe today’s lesson so far. The students could then move on to more elaborate communicative tasks, still centred around the grammar point.
Transfer-of-structure errors Since transfer-of-structure errors are caused by a contrast of rules in the native and target languages, the initial problem was to make the students aware of the contrasting grammar rules of Italian and English. It was obviously necessary to make direct comparisons with Italian. For instance, in attempting to eradicate errors in the use of the past participle, we used the following procedure: I We pointed out that such constructions as ‘My parents arrived, we went home’ and ‘The exam finished, I went to work’ were direct translations, and not acceptable in English. 2 We asked students to suggest alternatives. 3 When the students came up with such constructions as ‘When my parents had arrived, we went home’, we helped the students to develop hypotheses explaining the differences between Italian and English constructions. 4 We gave the students oral and written practice in using the correct form (e.g. skeleton sentences for the student to complete). In appropriate cases, particularly with advanced students, we went on to help students to develop their understanding of the relevant rules of English. Thus, in the example of the use of the past participle, we introduced the use of the past participle to replace a subject and a passive: He was awoken by the crash and jumped out of bed. Awoken by the crash, he jumped out of bed.
Overextension of analogy errors Here, we had to establish in the learners’ minds a distinction between Italian and English (e.g. the difference between It. ‘processo’ and Eng. ‘trial’), and also a distinction in their conception of English (e.g. the difference between Eng. ‘process’ and Eng. ‘trial’). However, we also had to deal with a large number of small separate problems, and not large-scale integrated problems, as was almost always the case with the other two error types. For this reason it seemed most appropriate not to try and eliminate these errors in one block of teaching, but to try and eradicate individual errors periodically in the course of the general teaching programme. One approach was periodically to give the students sentences to translate from Italian to English. These sentences included native language words that had given rise to misuse of the target language vocabulary. In addition, other exercises (e.g. reading and listening comprehension exercises) were peppered with the words the students had failed to produce correctly (in our sample ‘trial’, ‘job’, etc.). Finally, one could deliberately contrast the meaning of the confused target language words (e.g. ‘work’ and ‘job’), by including both words fairly close together in production and comprehension exercises.
Conclusion At a time when error analysis offers a considerable range of problems, both theoretical and practical, it is hoped that the essentially practical approach outlined here will be of use to teachers who are wondering how to deal with the errors their students make. • Received February 1982
References
Corder, S. P. 1973. Introducing Applied Linguistics London: Penguin. Dulay, H. C. and M. K. Burt. 1976. ‘Creative construction in second language learning and teaching’. Language Learning, special issue 4. McDonough, S. H. 1981. Psychologyin Foreign Language Teaching. London: George Allen and Unwin. Nickel, G. 1981. ‘Aspects of error analysis: “errare humanum est” ‘. AILA Bulletin 29 : 1-2 7. Selinker, L. 1972. ‘Interlanguage’. International Review of Applied Linguistics 1 0/3 : 201-3 1. The author
David Lott studied English Literature at University College London and was awarded a BA Hons degree in 1978. He now works as a ‘lettore’ at the University of Genoa in northern Italy, where he teaches English, History of the English language, and Psycho- linguistics.
ELT Journal Volume 37/3 July 1983