Contrastive analysis in language teaching, time to come in from the cold. (Language Teaching & Learning).

Abstract

The purpose of this short paper is to first argue that the rejection of the value of contrastive analysis (CA) input in foreign and second language teaching (FSLT) in the 70% and beyond was unjustified on both theoretical and empirical grounds and then, based on classroom-based research, demonstrate the value for teachers of using CA in their teaching.

**********

Very few teachers in FSLT of any experience would deny the value of some degree of explicit understanding of the grammar of the language they are learning. Though that understanding in some simple cases may be derived from exposure to comprehensible input, there is ample evidence to demonstrate that learners in normal classroom situations are unable to acquire most grammatical knowledge without the benefit of explicit grammar instruction. Furthermore, it is contended here that that understanding may be facilitated by an awareness of the differences between the L1 and the TL.

To put this in concrete terms, francophones when asked a question such as "How long have lived here for?" will tend to reply "I live here since two years", based on the French "Je demeure ici depuis deux ans." Similarly, they may say "I have 15 years" and "I am born in Montreal" calquing the French forms. To help learners overcome such problems, CA input enables them to understand the reasons for such errors and thereby go some way to avoiding them. In the case of the first example, one of the errors is caused by the fact that "depuis" may function with both the meaning of "for" and "since". Students need to understand this contrast and then to grasp the fact that "for" is used for a period of time whereas "since" refers to the point in time at which the period began as in "I have lived here four years/I have lived here since 1995".

All language groups engaged in FSLT will make many such L1-influenced errors. Furthermore, though all errors are potentially fossilisable, a number of studies have shown that it is L1-influenced errors which will prove to be the most persistent (See Sheen 1981, Marton 1981, Mukattash 1986). Given then the substantive presence of L1-influenced errors, their tendency to become fossilised, and, assuming the validity of the premise that learners need to understand the nature of that influence in order to overcome it, it is difficult to understand how the field of language teaching and applied linguistics allowed and even encouraged the rejection of CA input in the 70% and has continued to sanction it.

In order to help in the understanding of this rejection, I will trace the events which caused it and demonstrate the degree to which it was unjustified. Then, based on this and the description of classroom-based research studies, I will argue for the reintegration of CA input into FSLT.

Following the success of the Army Special Training Programme (partly based on behaviourist learning principles) in WW II in teaching American soldiers foreign languages, the early fifties saw the development in US universities of a structural teaching approach which was to develop into the audiolingual method (See Brookes 1964). This method revolutionalised and dominated language teaching in the decades following. CA became an integral part of the method as it was argued that habits associated with the L I would interfere with the learning of the TL. Thus, because francophones say "J'ai 15 ans", they will tend to say "I have 15 years" in English rather than "I'm 15", at least in the case of beginners. Thus. CA was exploited in order to identify those linguistic habits of the LI which were different from the TL and which might, therefore, cause errors. This then informed material writers as to the forms which would require special attention in the repetition and memorisation drills which characterized the audiolingual method.

Furthermore, it was argued that the large majority of errors were caused by L1 influence and that they could be predicted by CA. CA functioned, therefore, as an agent for prediction and as a means of providing linguiistic information to the material writers. However, it needs to be emphasized here that CA input was not used as a means of providing explicit instruction but as an indicator of which habits were to be eradicated by means of drills. As CA was then exploited to suit the purposes of audiolingualism, it became indelibly associated therewith and, therefore, was unjustly stigmatized as being essentially behaviorist whereas in actual fact it is neutral in terms of learning theory. That is it takes no position on the nature of the process of learning.

While audiolingualism was enjoying its success in the 60's, developments in theoretical linguistics inspired by Chomsky's work was to lead indirectly to its demise. Those developments were to result in the establishment of generative grammar as the received linguistic theory and the rejection of the behaviourist-inspired structural linguistics and, to cut a long story very short, in the acceptance of foreign and second language learning Not as the acquiring of a new set of habits but as a creative construction process (Kennedy & Holmes 1976; Krashen 1981) akin to the acquisition of the L1. That process was to be engendered by exposing learners to comprehensible input in communicative situations affording explicit instruction a minimal to non-existent role. Thus, audiolingualism died and was ultimately replaced by communicative-type methods.

Because of CA's link to audiolingualism and, therefore, behaviorist learning theory, it was viewed as being ill-suited to new communicative methods on theoretical grounds though this did not extend to practice in Europe (see Fisiak 1981). However, its ultimate rejection was also based on empirical arguments albeit ones of a spurious nature. The new field of error analysis became the means by which it was shown that many errors were caused by factors other than L1-influence (see Richards 1971, 1973). This was indeed true. However, as so often happens in this field, the tendency became exaggerated -- so exaggerated, in fact, that three prominent applied linguists of the period, Dulay, Burt and Krashen (1982), felt justified in making the following statement: Learners' first languages are no longer believed to interfere with their attempts to acquire a second language grammar, and language teachers no longer need to create special grammar lessons for students from each language background. (p. 5)

To support this, they cite Lance (1969), Richards (1971), Ervin Tripp (1970), George (1972), Valdman (1975), Hanania and Gradman (1977), all of whom arrive at very low estimates of grammatical L1-influenced errors. The Dulay et al. position is suspect for two reasons. First, it ignores other findings with appreciably higher estimates. Hocking (1973), Cornu (1973), Mougeon and Hebrard (1975), Sheen (1976), James (1980) and Steinbach (1981) arrived at findings which estimate the magnitude of such errors in the 60% range. Furthermore, as already mentioned, Marton (1981) and Sheen (1981) find that the errors in the English of near bilinguals largely comprise residual and possibly fossilized errors related to L1 influence. Second, a substantial number of studies involved in the body of method comparison research of the 60% and 70's had demonstrated the effectiveness of teaching methods using CA input. (see Von Elek & Oscarsson, 1973 for a review of this research and a description of their own rigorous work with a replication study). It is true that such research has been criticised for a lack of control of variables. Nevertheless, the consistency of the findings is such that they cannot be ignored.

In spite of this convincing counter evidence, it is the Dulay et al. position which has won the day. This can be seen by examining the text books for FSLT available today from publishers. They are all solely in English, making no reference to the L1 of the learners. This, of course, suits publishers for it enables them to sell the same text books all over the world thus increasing their sales many fold. It also suits the many anglophone teachers of English as a second or foreign language for it enables them to teach anywhere in the world without knowing the L1's of the students they teach. It might be legitimately argued, and for obvious reasons, that such text books are suitable for learning groups made up students with different L1's. However, most FSLT entails homogeneous L1 groups. Publishing solely monolingual texts without CA input and arguing that they are the best choice for all situations is, therefore, unjustified.

While this situation continues to characterize FSLT, developments in the field of second language acquisition research have refocused attention on the influence of the L1. The Gass and Selinker (1983) collection of papers was to reveal that "... that there is overwhelming evidence that language transfer is indeed a real and central phenomenon that must be considered in any full account of the second language acquisition process." (p. 7). That this was the case was later confirmed by the publication in 1986 of Kellerman's and Sharwood Smith's "Cross Linguistic Influence in Second Language Influence" and in 1989 Odlin's "Language Transfer".

In spite of these publications and others, there has been no apparent influence on the world of SFLT in terms of the availability of teaching texts which exploit CA input. This is perhaps understandable given the economic factors related to publishing and employment mentioned above and the theoretical bent of these publications. Publishing practices will not change unless economic pressures oblige publishers to do so. Such pressure needs to come both from the grass roots in the form of teachers' needs and from the demonstration by means of research studies of the efficacy of explicit CA input in SFLT. This has already happened to some degree in terms of a return to the inclusion of explicit grammar instruction content in text books. The early years of communicative methods saw the publication of text books which contained little to no explicit grammar content. However, as experience and research findings have demonstrated that the ignoring of some type of focus on form is not the optimal choice, publishers have returned to publishing text books with an underlying grammatical syllabus. This has not, unfortunately, extended to supporting it with CA input. Yet the acceptance of the principle of the need to understand the grammar of a language should justify the provision of CA input as it may facilitate that understanding.

Justification for this is provided in the already-mentioned method comparison research of the 60% and 70's. Following their review of that research and a description of their own replicated study, Von Elek and Oskarsson came to the following summary of their findings:

   The only safe conclusion one can draw is that, in the teaching of foreign
   grammar to adults, such techniques as grammatiical explanations, deductive
   presentations of the subject matter, translation, the use of the native
   language, and CONTRASTIVE ANALYSIS (my emphasis) are jointly superior to
   the combination of techniques constituting the implicit method.

It was this conclusion along with the related studies and my own practical experience as a teacher which led me to carry out my own research in order to compare a method based on communicative-cum-structural principles with a method founded on explicit teaching of grammar based on CA input where relevant. The findings are available in a paper published in 1996 in The International Review of Applied Linguistics (Sheen 1996). The study entailed comparing two comparable groups of Saudi Arabian graduate adults preparing for MBA study in the USA during a forty-week period of intensive English in Saudi Arabia. The results, operationalised in terms of frequency of error, are revealing for the study allowed for the comparing of the performance of the experimental group (EG) (ie with CA input) with the control group (CG) on those items involving CA input specifically. The items were as follows: 1. The copula (non auxiliary use of the verb "to be"). Arabic does not have it. 2. The determiners "a" and "an". Arabic does not have them. 3. The simple and prospective futures. Arabic only has the former. 4. The prepositions "in", "on", "at", "from". 5. The use of auxiliaries in question and negative forms. Arabic does not use them. 6. Relative clauses in which the relative pronoun refers to the object of the clause. In such cases, Arabic uses a resumptive pronoun as in "That's the car which I bought it". 7. The use of the verbs "take", "make" and "have". Arabic does not have the verb "have" as English does,

In the case of lessons with the EG entailing these items, they began with an explanation of how the two languages differed. Various exercises, including limited use of translation, were used to afford practice of the items both orally and in written form. Great emphasis was placed on the understanding of grammar and lexical meaning and the necessity for a conscious effort to commit it to memory and to use it correctly. In the the case of the CG, the CA items were given no special treatment but simply taught as were the other non-CA aspects, that is with a largely inductive approach but with the addition of grammatical summaries at the end of sections. In the case of the non-CA items, the EG received the same treatment as the CG. Thus the essential difference between the two groups was in the treatment of the CA items.

During the forty week period, there were five series of tests, eight weeks apart. Each series consisted of written, aural and oral tests. Subsequent statistical analysis by means of t-scores produced the following results. The EG performed significantly better on the CA items than did the CG in all three tests though the findings were the least significant in the oral parts. Thus, the treatment of the CA input was appreciably more Effective with the EG than with the CG in reducing the number of errors. It might be argued, however, that the difference may be accounted for by the fact that the greater emphasis on the CA items may have caused the significant difference. Equally, it might be argued that the EG had different teachers to the CG. However, if these arguments were valid one would expect that these differences would effect the results on the non-CA items. They apparently did not. The CG and the EG results for the non-CA items revealed no significant difference. It is, therefore, valid to conclude that it was the providing of CA input to the teaching of the CA items and the subsequent practice which produced the better performance on the part of the EG. This finding confirms other research already referred to; furthermore, research in the new field of language awareness provides further support for the advantage of providing CA input. (See Kupferberg & Olshtain 1995 and Trevise 1995).

Of course, one must be careful about generalising such findings to all teaching situations for it is not all learners who can benefit equally from explicit instruction. However, it does demonstrate the invalidity of the 70's absolute rejection of the usefulness of CA input in FSLT and though it may not result in publishers' accepting the usefulness of CA input and though it may not result immediately in a change in the policy of only producing monolingual text books, it is to be hoped that it will encourage those involved in FSLT to introduce CA input into their various classroom teaching strategies.

References

Brookes, N. (1964). Language and Language Learning. New York: Harcourt Brace.

Cornu, M. (1973). La France en direct, ler degre, au fil des lecons". Bulletin CILA 17, 30-47.

Dulay, H., M. Burt & S. Krashen (1982) Language Two. New York: OUP.

Ervin-Tripp, S. (1970). "Structure and process in language acquisition". In J. Alatis(ed., Monograph Series on Language and Linguistics, Washington D.C. Georgetown University Press.

Fisiak, J. (Ed.) 1981. Contrastive Linguistics and the Language Teacher. Oxford: Pergamon Press.

Gass, S. & L. Selinker(eds.) (1983). Language Transfer in Language Learning. Rowley, Mass.: Newbury House.

George, H.V. (1982). Common Errors in Language Learning -- Insights from English. Rowley, Mass.: Newbury House Publishers.

Hocking, B.D.W. (1973). Types of interference. In Focus on the Learner. T.W. Oiler & J.C. Richards (eds.) Rowley, Mass.: Newbury House.

James, C. (1980). Contrastive Analysis. London: Longman.

Kellerman, E. & M. Sharwood Smith (1986). Crosslinguistic Influence in Second Language Acquisition. Oxford: Pergamon Press.

Kennedy, G. & J. Holmes (1976). Discussion of creative construction in second language learning and teaching. In Dulay and Burr (eds.) 1976.

Krashen, S. (1981). Second Language Acquisition and Second Language Learning. Oxford: Pergamon Press.

Krashen, S. & P. Pon (1975). An error analysis of an advanced ESL learner. In Working Papers on Bilingualism, 7:125-129.

Kupferberg, I. & Olshtain, E. (1996) Explicit contrastive instruction facilitates the acquisition of L2 forms. Language Awareness, 5, 149-165.

Lance, D. (1969) A Brief Study of Spanish-English Bilingualism. Final Report Research Project Orr-Liberal Arts -- 15504. College Station, Texas A & M University.

Marton, W. (1981). Pedagogical Implications of Contrastive Studies. In Fisiak (ed.) (1981).

Mougeon, R. & P. Hebrard (1975). Acquisition et maitrise de l'anglais par lesjeunes bilingues de Welland. In Franco-Ontarian Section, OISE, mimeo.

Odlin, T. (1989) Language Transfer -- Cross-linguistic Influence in Language Learning. Cambridge:

Richards, J.C. (1971). A non-contrastive approach to error analysis. English Language Teaching, 25: 194-219.

Richards, J.C. (1973). Error analysis and second language strategies". In Oiler, J.W. & J.C. Richards (eds.) Focus on the Learner. Rowley, Mass.: Newbury House.

Sheen, R. (1976). L'analysedes erreurs commisesen anglais par des etudiants francophones au niveau universitaire. Unpublished doctoral thesis. Universite de Paris VII.

Sheen, R. (1979) Faux amis for francophones learning EFL. SPEAQ Journal Vol. 3. Nos 1 & 2:94-103.

Sheen, R. (1980). The importance of negative transfer in the speech of near-bilinguals". IRAL Vol XVIII/2:105-119.

Sheen, R. (1996) The advantage of exploitiing contrastive analysis in teachiing and learning a foreign language. International Review of Applied Linguistics. XXXIV/3 183-197.

Steinbach, H.R. (1981) On the classification of translation errors with some consideration of interference phenomena. In J. Fisiak (ed.) 1981.

Trevise, A. (1996) Contrastive metalinguistic representations: The case of "very French" learners of English. Language Awareness, 5, 188-195.

Von Elek, T. & Oskarsson, M. (1973). Teaching Foreign Language Grammar to Adults: A comparative study. Almquist & Wiksell: Stockholm.

Ron Sheen, University of Quebec, Canada

Ronald, Ph.D., is a professor of applied linguistics and English. He has taught in various parts of the world at all levels. His research interests are chiefly centred on methodological issues in second and foreign language teaching.