LANGUAGE DISTANCE AND THE MAGNITUDE OF THE LANGUAGE LEARNING TASK

S. P. Corder

(University of Edinburgh)

A learner brings to the classroom many characteristics which arerelevant to predictions about his career as a language learner. Thesecharacteristics are the product of his membership of a community; heshares its language and its attitudes to, beliefs about, motivations forand traditions in, language learning in general, and in the learning ofspecific second languages. And he possesses particular features of per-sonality as formed by his personal history of maturation and experience.I am concerned in this paper with the role of only two of these character-istics, both related to the community he belongs to: the nature of hismother tongue and any other languages known to him, and the beliefs cur-rent in his community, which he presumably shares, as to the nature, ex-tent and probable success in the learning task which lies ahead of him.

That members of a language community do hold certain beliefs aboutthe magnitude of the task of learning specific second languages and theirprobable sucress in doing so is, I think, incontestable. That they cangive some reasonable account of their assessment of the relative magnitudeof the task of learning different foreign languages is highly probable.The British Foreign Service, for example, quantifies this task by payingdifferent rates of language proficiency allowance, and also differentsubsidised periods of language instruction for its members, accordingto a scale of supposed or actually experienced difficulty in learningdifferent groups of languages. This scale appears to correspond to whatwe may call a measure of 'linguistic' or 'structural' distance of anyparticular language from English. Thus languages classed as meritingentitlement to the highest allowance include Japanese, Burmese, Chinese,Korean, those to an intermediate allowance, Polish, Russian, Persian andTurkish, whilst those attracting the lowest allowance include the usualEuropean languages, such as Danish, German and Spanish.

There are of course technical and theoretical problems in establishingand measuring degrees of language distance, but the assessment of thelearning task undoubtedly correlates with some notion of genetic related-ness as established by studies of language typology. Thus it is reportedand generally held that Persian is an easier language to acquire thanArabic, German than Russian, for native speakers of English.

Kellerman (1977, 1978) has been investigating the notions thatlearners have about the 'transferabi1ity1 of the forms of their mothertongue into second language performance. I would prefer to call thisphenomenon 'borrowabi1ity', for reasons given later in this paper. It

Corder - 28

appears that learners may have quite clear (but, of course, sometimes mis-taken) beliefs about what is similar in their mother tongue and the targetlanguage, and therefore 'borrowable1. In the earlier stages they seem toerr in regarding the mother tongue as more unique than it is in fact. Ashe progresses the more realistic his assessment of the uniqueness of hismother tongue becomes and the more willing he is to borrow from it, oras we might say, the more realistic his assessment of language distancebecomes.

I suggest, however, that the collective experience of a community oflearning different foreign languages does lead to a reasonably realisticassessment of the relative magnitude of the learning task of acquiring anyparticular foreign language, and that this largely corresponds to the for-mal linguistic relatedness of the languages in question to the mothertongue.

This notion is, of course, not a new one and the explanation for itwas held to be simple; the greater the degree of difference/distance, thelarger the learning task, or to put it another way, the longer the learningpath to be traversed between L, and l_2 (c.f. Lado, 19&1). This explana-tion, we may note, clearly assumed (a; that the starting point for learningwas the. mother tongue and (b) that the learning process was essentiallyone of restructuring (Corder, 1977a). The more restructuring required,the longer process, the greater the learning task. It is, of course,quite unnecessary to invoke any notions of comparative "difficulty" inthis explanation. The magnitude of a task is not logically connectedwith the measure of difficulty.

At this point we must narrow down the discussion to the acquisitionof the syntax of the target language. There does seem to be good evi-dence that the acquisition of much of the phonology of the second lan-guage is essentially a restructuring process. (Dickerson, 1975). No-one would seriously suggest that a second language learner replicates thephonological development of the infant acquiring its mother tongue. A-bout the acquisition of vocabulary I have no proposals to make, exceptto suggest that this is very probably highly contextually dependent asit clearly is in the case of the acquisition of the mother tongue vocabu-lary.

The hypothesis, then, that learning the syntax of a second languageis essentially a restructuring process was supported by the evidencethat, particularly in the early stages of learning, as might be predicted,a greater degree of mother tongue features in the learner's interlanguageperformance (hereinafter 'interference errors') are to be found. Thereare, however, two counter-arguments to this hypothesis. Firstly, by nomeans all learners show 'interference' errors in their speech and theamount of 'interference' varies considerably and unpredictably fromlearner to learner even under similar conditions of learning. Secondly,

Corder - 29

the restructuring hypothesis would predict that the learner's interlanguagegrammar remained at an equal level of complexity at all stages of learning.This is so manifestly not the case that the restructuring hypothesis can-not be sustained.

The alternative hypothesis first proposed by the author in 1967(Corder, 1967) and since investigated by many other workers (c.f. Dulay &Burt 1973, 1972*; Bailey, Madden & Krashen, 197*t; Larsen-Freeman 1975;Hatch & Wagner-Gough 1976; Hyltenstam, 1977) was that learners had a 'built-in syllabus1 for the learning of any particular second language. Thenatural cognitive processes of learning, when faced with a particular bodyof data, determined the sequence of creating that cognitive structurewhich we call the grammar of a language. The actual nature of the 'sylla-bus' or 'natural sequence', as it is sometimes called, was not at thattime specified, since no empirical studies of second language acquisitionhad then been made, but there can now be seen to be two possible versionsof the hypothesis of the 'built-in syllabus1. The strong version proposesthat all learners of a given second language follow roughly the same de-velopmental sequence whatever their mother tongue. The weak version pro-poses only that the developmental sequence is conditioned by the natureof the mother tongue.

There a> 2, in fact, two versions of the strong hypothesis: L| = !_2version (Dulay & Burt 1973) which claims that the developmental sequenceis similar for the learning of a language both as a mother tongue and asecond language, and another version which merely proposes that there isa particular 'natural sequence1 common to all second language learnerswithout specifying whether this is similar to the sequence of the mothertongue learner (Rosansky 1976).

The arguments against the strong version are that, if it is true ineither of its forms, then (a) any given L2 should present an equal learn-ing task, since it is assumed that whatever the route, the end points,fully complex natural languages, are equally complex.^ There is no evi-dence that natural languages vary in their overall complexity. Normalinfants acquiring their mother tongues all take much the same time toreach the same levels of communicative competence. Since it is manifestlynot the case that all languages do present the same learning task tospeakers of all mother tongues the strong version of the 'built-in syllabus'hypothesis appears to fail. (b) If the mother tongue plays no part inthe acquisition of the second language then one must find some alternativeexplanation for 'interference' phenomena in the learner's speech. This, Ibelieve, it is possible at least partially to do. See below. It does not,however, follow logically that if 'interference' phenomena are absent,the mother tongue (and other languages) play no role in the learning pro-ces. See also below.

Corder - 30

The arguments against the weak version of the 'built-in syllabus1

hypothesis are, however, equally cogent. (a) We have to explain away thenow very considerable evidence that learners, child and adult, free learn-ing and under formal instruction, do appear to go through sets of largelysimilar developmental sequences whatever their mother tongue (for a dis-cussion of relevant research see Krashen, 1977)• If the 'built-in sylla-buses' were indeed conditioned by the nature of the mother tongue then de-velopmental sequences would be predicted to differ considerably in thecases of learners having various, structurally different, mother tongues.This has not yet been shown to be the case. (b) The weak version of thehypothesis would also predict that "interference phenomena1 would besimilar and constant for all learners of particular mother tongues underall conditions of learning. This has also not been found to be the case(c.f. Clyne, 1968, Dulay & Burt, 1973).

It appears then that neither the strong nor the weak form of the'built-in syllabus' hypothesis (i.e. re-creation hypothesis, Corder, 1977a)can be sustained in its simple form. What is needed for an adequate modelof second language acquisition is a hypothesis which can reconcile the fol-lowing relevant findings;

1. The variability of the occurrence of 'interferencephenomena1 in differing learning conditions.

2. The considerable similarity of the sequential developmentof learners of different mother tongues when acquiringcertain aspects of some particular second language.

3. The relative different magnitudes of the task of learningdifferent second languages in relation to differentmother tongues of learners.

I shall take each of these points in turn.

1. The notion of 'negative transfer' has generally been invoked toaccount for the occurrence of 'interference phenomena' in interlanguagespeech. It was also allowed that the mother tongue might have a 'facili-tating' effect where Lj and l_2 systems resembled each other ('positivetransfer1). It was generally assumed that 'interference' or 'inhibition'and 'facilitation' were two sides of the same coin. In fact of course thisis not a logical necessity. The characteristics of the Li may 'facilitate'or 'not facilitate1 but 'failure to facilitate1 is not equivalent to'interfere' or 'inhibit1. It is perfectly logical to propose that the na-ture of the Lj may make passage along the built-in syllabus faster when itbears a similarity to L^, but simply has no effect when it is different.In such a case the learner is left with his own unaided cognitive learningcapacities to discover those aspects of the L£ which are not similar tohis L,. Facilitation clearly does not lead to 'interference phenomena1 in

Corder - 31

the learner's speech, but neither does non-faci1i tation. The errors madein either case will be developmental not interference errors; in the caseof facilitation, however, one may expect that the quantity of develop-mental error will be less since passage through the syllabus will beaccelerated. Where then do 'interference errors' come from?

An entirely plausible and commonsensical answer has long been avail-able. A learner when faced with the need to communicate will have recourseto whatever linguistic knowledge he has which will increase the likeli-hood of successful communication. (Corder, 1973). If he lacks therequisite knowledge of L9 to achieve successful communication then he willhave recourse to the L. or any other language he knows to make up thisdeficiency - beg, borrow or steal. And the greater the deficiency,typically at the beginning of a course, the greater the amount of borrow-ing (Taylor, 1975). As Newmark said: interference is simply the resultof a performer being "called upon to perform before he has learnt the newbehaviour. The result is 'padding1, using old knowledge, supplying whatis known to make up for what is not known". (Newmark, 1966, quoted byKrashen). And he adds "the cure for interference is simply the cure forignorance: learning". This is, of course, another way of saying thatthe motivation for the complexification of the interlanguage is the pres-sure of communicative needs. (c.f. Valdman, 1978).

It is now customary to make a distinction between 'learning strate-gies' and 'communication strategies'. 'Transfer' has usually been invokedas a learning strategy - the incorporation into the interlanguage grammarof mother tongue systems. We must remember here the original meaning of'transfer' in psychology - the persistence of, or resort to, already exist-ing behaviour in a functionally new behavioural activity. Let us also re-member that 'transfer' was a concept in behavioural learning theory and re-ferred specifically to what would now be called features of 'performance1.Competence or underlying cognitive structures have no place in such a theory.

The notion, however, of 'transfer' as a learning strategy would ac-count for the generation of sentences having mother tongue-like features,which, if the L, system were not identical with the l^, would be counted as'interference errors'. If similar, of course, transfer features wouldpass unnoticed. The alternative hypothesis outlined above treats 'inter-ference errors' as more or less ad-hoc unsuccessful borrowings to 'pad out'the gaps and inadequacies of the interlanguage system when faced with com-munication problems - that is, the result of a communication strategy.This account, of course, has two advantages: (a) It explains the varyingdegree of 'interference' found amongst learners with otherwise similarlearning conditions and knowledge, i.e. it relates 'borrowing' firmly tocommunicative needs and actual performance and (b) it accounts for therelative consistency of the types of unsuccessful borrowing (interferenceerrors) and relates the known or predictable gaps in the 'built-in syllabus'at various stages to the available 'paddings' from the L, or other known

Corder - 32

languages. The more demanding the communicative activity required of alearner (i.e. the greater the mismatch between 'means' and 'ends', 'know-ledge' and 'need') the greater the amount of borrowing in general and henceof unsuccessful borrowing (interference error).

We may note here that 'interference errors' are strongly associatedwith classroom language activity (I will not dignify it with the title ofcommunication) and this is what would be predicted where it is not thelearner himself who controls the nature of this activity. In a free com-munication situation a learner can adopt all sorts of alternative strate-gies to 'borrowing' which are generally deprecated or banned in the class-room, e.g. gesture, guessing, periphrasis, semantic avoidance, etc.(Tarone et al. 1976).

2. The second finding we have to account for in the theory is thestrong similarity observed in the sequential development of some aspectsof learners' interlanguage from differing mother tongues (c.f. Hyltenstam,1977) I have elsewhere (Corder, 1977c, 1977b) suggested (and I will notpresent the arguments here) that the starting point of the language learn-ing continuum is a basic, simple, possibly universal grammar to which alllanguage users have access, because in the process of language acquisitionevery infant creates such a simple grammar. We have access to this grammarbecause we do not forget our own linguistic development; we do not kickaway the ladder by which we climb, but are able under certain circumstancesto utilise such grammars for communicative purposes as adults (e.g.'foreigner talk1, 'baby talk', etc.) and also for second language acquisi-tion (c.f. Traugott, 1973, Ervin-Tripp, 197*0 • It cannot be an accidentthat the early stages of both Li acquisition, 1-2 acquisition, 'simplifiedregisters' and pidgins all show striking formal similarities. This hy-pothesis has already received considerable support from the studies ofmorpheme acquisition (Schumann, 197*0 in adults and children in secondlanguage acquisition.

3. Thirdly, what we must also account for is the relative dif-ferent speed of learning different second languages in relation tovarious mother tongues. As has been suggested, the time taken, or thelearning task faced, by an infant acquiring any language is assumed to beof approximately equal magnitude. All infants achieve similar levels ofcommunicative competence at approximately the same stage of development.If then second language learners are engaged upon the task of creatingfor themselves a grammar of any particular second language, all startingfrom the same point, and apparently all following the same developmentalsequence thereafter, why is it that they typically take different timesover the job? Why is it that apparently the same task differs in magni-tude for different groups of learners? The hypothesis here proposed statesthat, other things being equal, (e.g. motivation and access to data, etc.)the mother tongue acts differentially as a facilitating agency. Where themother tongue is formally similar to the target language the learner willpass more rapidly along the developmental continuum (or some parts of it),than where it differs. Genetically related languages are assumed to share

Corder - 33

a larger number of rules, particularly in the deep grammar, differingprincipally in the more superficial aspects. Passage along the develop-mental continuum is therefore rapid until those relatively superficialdistinctions are met, whereas in the case of unrelated (distant) lan-guages differences exist along the whole developmental continuum, slowingdown the speed of acquisition. This hypothesis is testable by a compara-tive study of learners acquiring two different languages simultaneouslyunder the same learning conditions, e.g., exposure, teaching, motivation,etc.

We must remember here that not only the mother tongue may be facili-tative. Other languages known to the learner, however imperfectly, may,in the degree to which they resemble the target language structurally,have a facilitating effect. This hypothesis is supported by the verygenera] observation that the more languages one knows the easier the ac-quisition of yet another appears to be. This is usually explained by thenotion that in such cases the learner has a larger number of 'ready-made'hypotheses to test in processing the data of the new language. Let usnote here in parentheses that the possession of a number of languages mayactually act as a facilitating agency in the manner just suggested butalso provide an increased resource for 'borrowing' as a communicationstrategy.

In summary, the model of the learning process that emerges is one inwhich the learner starts his learning programme from a basic, possibly uni-versal grammar which he proceeds to elaborate in response to his exposureto the data of the target language and his communicative needs. The ela-boration follows a more or less constant sequence for all learners of aparticular language but any particular learner's progress along the devel-opmental continuum is significantly affected by the degree to which hisexisting knowledge of language may facilitate his advance. This facilita-tion does not manifest itself in the transfer of mother tongue features tohis interlanguage grammar but in the more rapid discovery of the mother-tongue-like features of the target language. Where unacceptable mother-tongut-like features appear in his speech they are the result of an un-successful communication strategy of 'borrowing1. All of us, in otherwords, already know the second language to a greater or lesser extent.Part of the task of acquiring a second language is finding out how muchwe already know of it. The more we find we know, the less the magnitudeof the learning task.

Corder - ?>k

NOTES

1. The expression 'learning task1 is used here to refer to 'what has tobe learned1, and has been chosen to avoid any implication of relativeease or difficulty of learning. The stereotype beliefs in any com-munity about the learning task will presumably be a compound of no-tions about the work to be done to achieve, for example, some levelof communicative competence in the language, including culturalknowledge, some level of acceptable and functionally adequate pro-nunciation and some level of syntactic control. For any particularlanguage one or another aspect of the language or its use may bemore dominant. The discussion in this paper will concentrate on thetask of learning the syntactic system of the language.

2. One must, of course, here concede that the value to the Service mayalso play a part in the assessment of proficiency allowances, i.e.They must also act as incentives.

3. It is here assumed that the learning objectives are necessarily thefull standard language. In certain learning situations this may notbe the case. (c.f. Platt, 1977; Pienemann, forthcoming).

h. Where we have two different languages which are virtually so close asto constitute 'dialects', e.g. Swedish and Norwegian, we may have topostulate that we are not concerned with language acquisition at all,but something we could call 'dialect expansion'. In such cases thelearner already knows the target language and is concerned merely withdiscovering the purely superficial and trivial differences between hisdialect and the target dialect. This proposal sidesteps the problemof determining when dialect distance becomes so great that we have todo with different 'languages'. But see below for a possible explana-tion.

Corder - 35

REFERENCES

Bailey, N., Madden, C. & Krashen, S. 1974. Is there a Natural Sequencein Adult Second Language Acquisition? Language Learning, 24.235"243-

Clyne, M. 1968. Zum Pidgin-Deutsch der Gastarbeiter. Zeitschrift furMundartforschung, 35. 130-139•

Corder, S.P. I967. The Significance of Learners' Errors, IRAL 5.4. 161-70.

1973. Introducing Applied Linguistics, Penguin: Harmonds-worth.

1977a. Language Continua and the Interlanguage Hypothesis.In Corder & Roulet (eds.) Proceedings of the Vth Neuchatel Colloquium,Librairie Droz: Geneva.

1977b. Simple Codes and the Source of the Second LanguageLearner's Initial Heuristic Hypothesis. Studies in Second LanguageAcquisition, 1 (1) . 1 -11 .

1977c. The Language of Kehaar. RELC Journal, 8.2. 9-35.

Dickerson, L.J. 1975- The Learner's Interlanguage as a System of VariableRules, TESOL Quarterly,9.4.401.

Dulay, H. & Burt, M. 1973. Should we teach Children Syntax? Language Learning,23.2.245-258.

1974. Natural Sequences in Child Second Language Acquisition,Language Learning, 24.1.37-53.

Ervin-Tripp, S. 1974. Is Second Language Learning like the First? TESOLQuarterly, 8.2.

Hatch, E. & Wagner-Gough, J. 1976. Explaining Sequence and Variation inSecond Language Acquisition. In Brown, D. (ed.) Papers in SecondLanguage Acquisition, Special Issue No. 4 Language LearnTng"

Hyltenstam, K. 1977- Implicational Patterns in Interlanguage SyntaxVariation, Language Learning, 27.2.383-411.

Kellerman, E. 1977- Towards the Characterization of the Strategy ofTransfer in Second Language Learning, I.S.B. 2.1. 58-146.

1978. Giving the Learners a Break; Natural Language Insti-tutions as a Source of Predictions about Transferabi1ity, WPB 15-59-93.

Krashen, S. 1977- Some Issues relating to the Monitor Model. On TESOL77, TESOL: Washington.

Corder - 36

Lado, R. 1961. Linguistics across Cultures, University of Michigan Press:Ann Arbor.

Larsen-Freeman, D. T975- The Acquisition of Grammatical Morphemes byAdult Second Language Students, TESOL Quarterly 9.4. 409-419-

Newmark, L. I966. How not to Interfere with Language Learning, InLanguage Learning; the Individual and the Process, Int > Rev. AmericanLinguistics, 40. 77-83.

Pienemann, M. (forthcoming). Erwerbsequenzen und Lehrprogression. Uber-legungen zur Steuerung des Zwei tsprachenwerbs ausla'ndischer Arbeit-erkinder, Forschungsgruppe Wuppertal.

Platt, J. 1977. The 'Creoloid' as a Special Type of Interlanguage, I.S.B.2.3. 22-39.

Rosansky, E. 1976. Methods and Morphemes in Second Language Acquisition,Language Learning, 26.409~425-

Schumann, J.H. 1974. The Implications of Pidginisation and Creolisationfor the Study of Adult Second Language Acquisition, TESOL Quarterly,8.2. 1*6 152.

Tarone, E., Cohen, A.D. & Dumas, G. 1976. A Closer Look at some Inter-language Terminology; A Framework for Communication Strategies.Working Papers in Bi1ingualism, 9. 76-91-

Taylor, B. 1975. The Use of Overgeneralization and Transfer LearningStrategies by Elementary and Intermediate Students in ESL, LanguageLearning, 25-73-107.

Traugott, E. 1973- Some thoughts on Natural Syntactic Processes. InBailey, and Shuy, R.W. (eds.) New Ways of Analysing Variation in En-glish, Georgetown University Press: Washington.

Vaidman, A. 1978. On the Relevance of the Pidginisation-CreolisationModel for Second Language Learning, Proceedings of the Vlth NeuchatelColloquium 1977, Studies in Second Language Acquisition (2)55-75-