Contrastive Analysis

The significance and role of the L₁ in FL syllabi – a historical overview

We dissect nature along lines laid down by our native language…
Language is not simply a reporting device for experience but a defining framework for it.
(Benjamin Lee Whorf 1936; cf. also 1940/56:212)

The use of the learners’ native tongue in the language classroom is nothing new; since the embryonic stages of what could be called regular education it has probably been the most natural and commonsensical approach. At the beginnings of the past century, Harold Palmer observed: “[t]he first and foremost qualification of the ideal teacher is a thorough knowledge of both the foreign language and the student’s native tongue” (Palmer 1917). Over the years the learners’ L₁ has been utilised in a gamut of purposes: to convey meaning, dissect language, teach grammar, explain errors, discuss cross-cultural issues, assess comprehension, give instruction, and manage the classroom (Atkinson 1987; Auerbach 1993; Schweers 1999; Cook 2001b; Deller & Rinvolucri 2002; Schultz et al. 2002).

Yet, the place and use of the L₁ in the L₂ classroom (both as a medium of instruction and as tertium comparationis – the constant in comparison) was frowned upon and advised against during a good portion of the past century: “It is assumed throughout that the teacher’s success is judged by the rarity of his lapses into the foreign tongue” (Thorley 1910). There has been a long tradition in the métier of ELT of not only ignoring the mother tongue in the learner’s mind, but also of suppressing its surfacing in the classroom, and many teachers still feel guilty if the learners’ L₁ was let in. Similarly, the overwhelming majority of language course books and grammar reference materials on the market (with a few notable exceptions where contrastive grammar boxes are present^[1]) provide monolithic, one-size-fits-all English-language explanations and totally ignore the relations obtaining between the students’ L₁ and the TL, forgetting the observation made by Canadian poet Robert Zend that “[p]eople have one thing in common: they are all different.” As Cook (2001b) puts it, the writers have adopted the 19^th-century injunction to eschew the first language as much as possible in the classroom rather than seeing it as a resource for teaching:

As a rule, at the beginning of the twenty-first century, language teaching classrooms are still supposed to emphasise the spoken language, to avoid the first language and grammatical explanation and to practice language in whole dialogues rather than in isolated fragments – however different actual classroom practice may be. (Cook 2002b:329)

Such mainly Euro- or Amerocentric books moulded in the generic approach are, using James’ (1980:24) term, “universally valid [but] for purely commercial reasons.” In the words of Howatt (1984:289), “the monolingual principle, the unique contribution of the twentieth century to classroom language teaching, remains the bedrock notion from which the others ultimately derive.”^[2] Even today in numerous classes across the globe the L₁ is banned completely and all activity essentially resembles the ‘Direct Method’, where all communication is supposed to take place in English, or through gesture and mime when the former fails (Grant 1993:v).

Nevertheless, there are times when enormous advantages lie in selectively and sparingly using the L₁, “an acknowledgement that many teachers will welcome” (Grant 1993:v), and the L₁ has fortunately never lost support in the field (cf. e.g. Atkinson 1987; Harbord 1992; Lightbown & Spada 1999; Schweer 1999). As Majer (2006) writes:

A general conclusion ensuing from the analysis of extensive material, ranging from mere lexical gaps of the beginner learner to deliberate and conscious playful juxtaposition of L1 and L2 of the advanced speaker, based on both self-reported transcripts of classroom talk and samples of naturalistic multilingual speech reported in the literature, is that foreign language instruction is essentially a bilingual methodology organised predominantly around concurrent allocation of the two languages rather than their clear separation.

Currently the importance of the L₁ is becoming a matter of received wisdom (Arabski 1996:31), or at least a persistent issue. After all, the native language “plays a part at the start of learning, in the process of learning, and in the use of the target language in communication” (Corder 1993:29).

One of the major utilities of the L₁ is the expediency of cross-linguistic comparisons. As Lado (1957) rightly observed, “[t]he teacher who has made a comparison of the foreign language with the native language of the students will know better what the real problems are and can provide for teaching them.” Noticing the most problematic contrasts between the L₁^[3] and the TL and helping learners overcome the arising difficulties facilitates and accelerates the learning process, making teaching in monolingual classes in particular extremely efficient (Atkinson 1993:8). Native speakers’ grammars are “-etic or ‘externally grounded’ whereas they ought to be -emic or ‘congruent with the point of view of the individual being investigated’” (James 1994:208), and learners need to be taught grammar from their own perspective. This necessity was duly recognised over four decades ago as central to success in language teaching:

There is no short cut to learning how to teach English as a second language effectively, but it helps very much if the teacher can begin to be aware of the language from the point of view of his pupils. He needs to become aware of English … in an entirely new way. … This language awareness is the basis of all the teacher’s work. (Derrick 1966)

Without being cognizant of the areas of interference, teachers may fail to perceive the cause of or understand learners’ erroneous production. Having a knowledge of—and sensitivity to—the ways in which the learners’ L₁ and L₂ differ “is beneficial both as an aid to localising areas of potential difficulty for the learner where interference may occur and as a vehicle for explanation, in giving learners feedback on their own speaking and writing” (Leech 1994:21). The teacher will then be able to make informed choices concerning selection and emphasis, knowing which aspects of the TL to concentrate on and which can safely be left aside. The ideal solution for a Polish learner of English would be a textbook written by a Polish author cognizant of the areas of potential difficulty as well as those wherein positive transfer can be invoked^[4] – already in 1968 Hall asserted that the era of the uniform, standard textbook for all learners of a FL, irrespective of their language background, is over: the whole composition of the course book, from the selection of items, through the degree of prominence accorded to each, the nature of exposition, and exercise formats, should be customized with their L₁ (Sridhar 1980/81:212). The students themselves will, after all, resort to their L₁ anyway. Knowledge of the mother tongue serves as a springboard for FL development as “interlingual associations … provide a key to the mysteries of the FL” (James 1994a:212). Also Selinker (1992:171) puts forward the argument that learners frequently fall back on their knowledge of the L₁ when they encounter a new linguistic form in the input.^[5] Nowadays numerous scholars argue that a cross-linguistic approach promotes raising learners’ consciousness of interlingual similarities, and accentuate the crucial importance of the students building not merely a habit, but a foundation in knowledge (Pratt-Johnson 2006:3). Last, but not least, “success in learning a foreign language is contingent on a certain degree of maturity in the native language” – this observation made by Vygotsky as early as in 1934 (1962:121) is only recently being recognised as a novel discovery by numerous researchers (cf. also Sections 1.4, 4.2.3, Supplements K.7 & K.14). The saga of contrastive language studies will be compacted into the next four sections, so that when the Reader reaches Chapter 4, the Language Interface Model can be seen as taking up the best of the (good) practice.

Contrastive Linguistics – a historical perspective

The history of contrastive linguistics (CL, otherwise known as ‘contrastive analysis’—CA—or ‘contrastive studies’—CS) has had the wheel-of-fortune character (as aptly presented in Fisiak 1981 and Granger 2003:17f.) and its status, validity, and pedagogical application have been the topic of innumerable spirited controversies since the mid-sixties. For a long time it was believed to be central to all linguistic research (Ferguson 1968; Jackson 1976):

-          in developing a general theory of language basing on the discovery of universals,
-          in the study of diachronic change—comparisons of various stages in the historical evolution of a language,
-          in the study of dialectal variation,
-          in confronting different languages synchronically and charting their typology,
-          in contrasting two or more languages or their subsystems in order to establish similarities and differences,
-          in longitudinal studies of language acquisition,
-          and in translation.

Comparative studies in the métier of linguistics are no recent phenomenon. The first systematic theories of the liaisons between human languages, the beginnings of modern philology, Indo-European studies and—by extension—of contrastive linguistics were marked by English judge, Sir William Jones’^[6] (‘Oriental Jones’’) justly famous Anniversary Discourse of Feb. 2 1786 (published two years later; Jones 1788), where he contended that Sanskrit must be related to Latin and Greek, and surmised that Germanic and Celtic descended from the same source “though blended with a very different idiom.” Allow me to quote his famed ‘philologer’ passage:

The Sanscrit language, whatever be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either, yet bearing to both of them a stronger affinity, both in the roots of verbs and the forms of grammar, than could possibly have been produced by accident; so strong indeed, that no philologer could examine them all three, without believing them to have sprung from some common source, which, perhaps, no longer exists.

Although the term ‘contrastive linguistics’ dates to Whorf (1941), the roots of more voluminous research extend at least to the turn of the 20^th century, with most of the first studies predominantly theoretical (one may mention here the such contrastive analyses ante litteram as Grandgent 1892; Viëtor 1894; Baudouin de Courtenay 1912; Passy 1912; or Bogorodickij 1895/1915), the applied aspect (e.g. Viëtor 1903) definitely more tangential and of secondary importance (Fisiak 1981:4). This theoretical strand was perpetuated well into the sixties by linguistis of the Prague School, notably Vilém Mathesius (1928, 1936) and his followers (Bohumil Trnka 1953-5, Alexandr V. Isačenko 1954-60, Josef Vachek 1961, and Jan Firbas 1964).

Across the Atlantic, great interest in CS was stimulated by WW2; it was recognised as a vital part of language teaching methodology and granted huge funds, following the declaration by Charles C. Fries that:

the most efficient materials are those that are based upon a scientific description of the language to be learned, carefully compared with a parallel description of the native language of the learner. (Fries 1945:9)

Modern CA—whose beginnings are thus frequently traced to the forties (Fries 1945; Trager 1949) and fifties, when it was established as an integral component of language teaching methodology (Sridhar 1980/81:209)—developed out of the contemporaneous behaviourist-inclined SLA and FLL philosophies and programs, borrowing the notions of ‘transfer’ and ‘interference’ from psychological learning theory (Jackson 1981:195). Behaviourist views of language learning prevailed in the two decades following WWII, drawing on the general psychological theories of learning propounded by e.g. Burrhus Frederic Skinner (1957), the underlying assumption being that LL—any learning, for that matter—takes the form of habit formation, inductively through analogy rather than analysis (Ellis 1994:300). From this perspective, interference from prior knowledge—or proactive inhibition, when old habits get in the way of attempts to form new ones—was taken to constitute the main impediment to learning (cf. Section 1.11.3). Consequently, the degree of difficulty in learning was believed to mirror the extent to which the TL pattern differs from the NL one, and there was a tendency to bring down to interference all TL errors which showed similarity to an L₁ feature. Errors were hence considered detrimental to successful learning, preventing the formation of correct habits, which sometimes resulted in a pathological errorophobia.

CA was initially developed with the traditional FL classroom in mind, with English as the L₁ of the students and the teacher, and one of the other European languages as the target (Jackson 1981:198). It also tended to focus primarily on phonological systems. Originally a purely applied enterprise, it aimed at developing more efficient FLT tools and methods. A deluge of pedagogically-motivated contrastive these, dissertations, papers and monographs appeared (dissecting e.g. German, Spanish, Italian, French, and Russian; ibid.), chiefly under the auspices of the Center for Applied Linguistics in Washington, DC, with the aim of predicting and discovering learner difficulties. For a long time, these studies remained underlined by the naïve, yet highly influential credo of the prime movers of Contrastive Analysis Hypothesis (CAH)—laid out by Weinreich (1953) and Lado (1957) in his seminal field manual and prevalent well into the sixties—that the degree of similarity between items in the NL and the TL stays in inverse proportion to the difficulty involved in learning these in the process of SLA:

the student who comes in contact with a foreign language will find some features of it quite easy and others extremely difficult. Those elements that are similar to his native language will be simpler for him, and those elements that are different will be difficult. (Lado 1957:2)

This was partly due to the traditional preoccupation of CA with populations of learners sharing the same L₁ (James 1979/81:63), so that errors were traceable to that source. Starting from the Ansatz that difference translates into difficulty in learning, Contrastive Analysis boiled down to charting areas of similarities and differences between languages, to be later considered in syllabus design. These attempts resulted in hierarchies of difficulties^[7] (e.g. Stockwell et al. 1965; Stockwell & Bowen 1965), which were often expected to automatically provide the optimal solution in syllabus sequencing. A spate of horizontally organised contrastive analyses of major European languages followed with the boom of the CAH in the 1960s, primarily designed to assist course developers in the preparation of teaching materials, resulting in inventories of similarities and differences and culminating in the publication of the Contrastive Structure Series volumes in 1962-65 (op. cit.).^[8] At the same time, contrastive linguistic research not only highlighted differences between the languages in question, but also revealed and enabled to trace details and subtleties in each of the languages that might have gone unnoticed but for the contrast with another language (Aijmer & Altenberg 1996:12; Aarts 1998; Johansson 2000:4), especially in the case of closely related languages with nuanced differences (Rawoens 2006:2).

Fisiak (1981:2) distinguishes two main strands in the saga of contrastive linguistics:

·       theoretical, with the desire to contribute to a) the establishment of linguistic universals, and b) increasing detailed knowledge of particular languages (Sharwood Smith 1974/81). Conducted along horizontal dimensions, it sought to systematically compare two—or more—languages with the aim of offering an exhaustive account of the differences and similarities between them, providing an adequate model for their comparison, and defining how and which elements are comparable, hence (Krzeszowski 1977/81:71f.) centring around morphosyntactic considerations associated with the notion of congruence (Krzeszowski 1967; W. Marton 1968), correspondence, and semantic considerations associated with the crucial notion of equivalence (Halliday et al. 1964; Catford 1965; Krzeszowski 1967)—the necessary criterion for comparability. It consisted in isolating systems, subsystems, constructions, or rules for comparisons with their cross-linguistic equivalents, resulting in descriptions of their presence or absence, obligatory or optional status, and hierarchy in the respective languages. Exploring how universal categories are realised in the contrasted languages, theoretical contrastive studies were thus frequently language-independent and non-directional;

·       applied, with aims related to language teaching and the development of teaching materials (Sharwood Smith 1974/81). Drawing on the findings of theoretical CA and providing a framework for the comparisons, it selected information pertinent for the purpose (teaching, studies on bilingualism, translation, etc.) and presented it adequately. As such, it devoted more attention to surface representations than theoretical contrastive studies did, since these are what the learners or translators have more immediate access to and what language teaching has always been concerned with. It was unidirectional, investigating how a (universal) category realised in one language in one way is rendered in the other and what the consequences for a given field of application may be, and identifying potential areas of difficulty due to interference. Vitally, the value of pointing out similarities in language teaching was emphasised, with the aim of relieving the learner from having to guess them and preventing him/her from attempting to construct forms which may ‘sound foreign’ (Fisiak 1981:3).

Prof. Jerzy Zybert (p.c. Oct. 13 2006) adds that the aim of developing pedagogical materials was more visible in the US, while Europe was more interested in the theoretical dimension.

Criticism of CA

In its heyday, the value of such taxonomic CA went virtually unquestioned (Sanders 1976/81:21) and CS were regarded as a panaceum for all the aporias in language teaching (W. Marton 1972/81:157). Yet, some cracks soon appeared in the picture. Lado’s (1957:vii) dictum to the effect that patterns which will or will not cause difficulty can be predicted and described “by comparing systematically the language and culture to be learned with the native language and the culture of the student” was one of the primary causes of the controversy which ensued in the 1960s (Sajavaara 1981a:34). The simplistic and restrictive CAH and the claims that a contrastive description would successfully predict all learner errors fell out of favour in the 1970s. The hopes and expectations of rapid advancements in language teaching were very precipitate, and it soon became clear that CA could not deliver the goods expected (Jackson 1976). A weak version of the CAH was proposed by Wardhaugh (1970), postulating the use of CA a posteriori hand in hand with error analysis to explain rather than predict (cf. also Ritchie 1967, 1971; Lee 1968; Gradman 1971; Whitman & Jackson 1972, inter alios). However, this also turned out to be impractical and inadequate: James (1980) argued that undertaking a time-consuming comparison simply to confirm transfer as the suspected cause of errors is not worthwhile; Schachter (1974) additionally emphasised the need to identify areas of possible avoidance. Other factors turned the scales in CA’s disfavour, on both external (empirical validity of the predictions made) and internal grounds (the theoretical basis):

·       methodological (ir)relevance for language teaching and the abstract nature of the analysis (Sajavaara 1981a:39, 43)

The contrasts produced by CA were mostly so complex, abstract, and full of highly technical detail that their pedagogic application seemed nil;

·       the predictive nature of CA

More often than not, most of the results produced by traditional CA were platitudes known to every experienced language educator (Stevens 1970), claimed to be easily attainable by means of error analysis. By the end of the seventies the assumption additionally came under criticism following a body of SLA research where the role of intralingual mechanisms such as overgeneralisation of target rules was found to be equally—if not more—influential in language learning. The strong version of the CAH stated in Lee (1968:186)—which failed to acknowledge sources of difficulty other than the learner’s L₁ (e.g. lack of saliency, low communicative value, markedness, processing difficulty) and took no account of the psychological processes^[9] involved in language learning—became theoretically untenable with the realisation that grammatical deviance is not always nor exclusively proportionate to the degree of mismatch between L₁ and TL forms (Whitman & Jackson 1972). L₁ influence is only one of many sources of error (others being caused e.g. by overgeneralisation, or inconsistencies in the spelling system of the TL). At the same time, a large portion of the predictions soon became invalidated (Brière 1968; Politzer 1968): many errors occur that are neither traceable to transfer (Dulay & Burt 1974a) nor illuminated by CA, while many of those overpredicted by the CAH—including transfer load—do not show up in actual performance at all (Jackson & Whitman 1971). Differences in particular areas of language systems only cause interference in some cases; psychological, pedagogical, and other extralinguistic factors—unaccounted for by CA—also contribute to the formation of errors. CAH also disregarded the fact that the operation of transfer is heavily influenced by individual factors such as learning strategies^[10] employed (e.g. Slama-Cazacu 1971; Nemser & Slama-Cazacu 1970). Hence, the issue of the predictability of interference and errors was raised against CS (Fisiak 1981:7), which had failed to meet the criteria of both descriptive and explanatory adequacy^[11]. This resulted in questioning the relevance of CA in designing FL instruction materials (Krzeszowski 1977/81:73);

·       association with behaviourism

The rejection of transfer as a major factor in SLA was also an overreaction engendered by the close affinity between the concept and the behaviourist conditioning principle it had been based on (Dulay & Burt 1972, inter alios), which fell on evil days (Sanders 1976/81:21), primed by Noam Chomsky’s (1959) review of Skinner’s work. Crucially, it was thundered that learning by analogy cannot account for learners’ generative potential, and that parents rarely correct their children’s ill-formed utterances (the No Negative Evidence hypothesis), suggesting a developmental make-up of language acquisition (and of certain error types which occur in the ILs of all acquirers). Regrettably, the demise of behaviourism brought with it a neglect of the role of cross-linguistic influence, throwing the baby out with the bathwater: “In clambering on to the mentalist bandwagon … researchers like Dulay and Burt mistakenly dismissed transfer, often on the basis of flimsy evidence” (Ellis 1994:315);

·       structuralism, turmoil in the theory of grammar, and the independence of individual linguistic descriptions (Sajavaara 1981a:39-42)

The incipient phases of taxonomic CA were closely meshed with American structuralism as advocated by Edward Sapir and Benjamin Whorf, and many analyses were conducted under the influence of the Bloomfieldian type of structuralism with its insistence on differences between languages and concentration on language-specific features of autonomous, self-contained linguistic systems. Consequently, no complete description of any two languages according to one and the same model was developed, as these were created in close adherence to descriptions of certain individual languages (James 1976), rendering the independence of such fragmentary descriptions questionable. Basing the early stages of modern CL on surface ‘American structuralist’ descriptions was another obvious cause for later criticism (Sajavaara 1981a:35);

·       preoccupation with the surface structure (Sridhar 1980/81:213)

With the advent of transformational grammar, taxonomic CA—like taxonomic descriptive linguistics in general—became disparaged for its preoccupation with the surface structure of language (cf. Di Pietro 1968, 1971).^[12]

·       fragmentariness

Coupled with the fact that they mostly constituted only fragmentary descriptions of the language systems, this sometimes gave rise to fears that basing teaching materials on CAs may result in the learners’ being presented with only bits and pieces of the TL system (W. Marton 1972/81:161);

·       syllabus choice and sequencing factors

Those who expected hierarchies of difficulty to automatically provide the best indicators for syllabus sequencing disregarded such factors as the relative, immediate communicative expediency of the target items, their frequency of occurrence within the relevant style/register, simplicity, the economy of their alignment with other structures, or motivational strength (W. Marton 1972/81:164; Jackson 1981:196);

·       the methodological problem of equivalence (Sajavaara 1981a:41)

CA was primarily developed basing on translational equivalence as established by a bilingual informant; a rather unreliable method for the subsequent establishing of the surface categories corresponding to certain deep semantic entities (James 1976);

·       the theory of transfer (Sajavaara 1981a:41f.)

Under the influence of Harris (1954) and the psychology of learning, the theory of transfer (i.e., the hypothesis that the learning of a task is either facilitated or impeded by the previous learning of another) played an important role in the development of CA and EA. However, it proved rather difficult to tell which kind of items are susceptible and at which level this transfer takes place (Slama-Cazacu 1971; James 1976). Moreover, it was difficult to draw a boundary between interference proper and various unconscious learning strategies (Kellerman 1977);

·       the static nature of the analysis

CA regarded the learner’s position in relation to both the TL and the L₁ as stable, failing to perceive radical changes in the learner’s stand in relation to both languages with the increase of FL operationality: an elementary learner is in a different position than an advanced one, a child’s position is different from that of an adult (Sajavaara 1981a:44). Consequently, proponents of EA renounced the one-sided description of learner errors in favour of the study of IL, replacing the rather static basis of analysis with one of the language learning process, where the IL is no longer seen as an erroneous form of the TL, but an état de dialecte (Corder 1972, 1975b);

·       teacher-centeredness

Newmark and Reibel (1968:149) attacked CA as endorsing a teacher-centred rather than a leaner-centred approach to FLL;

·       misapplication of theoretical analyses

CS elaborations not intended for classroom consumption were sometimes believed otherwise, and following this misapplication, much too abstract and unable to attain the overambitious goals for which they were not designed in the first place, they consequently also suffered for the failures of CL (Slama-Cazacu 1971; Bausch 1973; Fisiak 1981:5);

The numerous misinterpretations and misunderstandings created by the peculiar methodological status of CS, the lack of a clear-cut distinction between the theoretical and applied branches and of a precise formulation of their different aims in the past (Stockwell 1968; Fisiak 1971), additionally exacerbated by a confusion of the relationship between CS, the psycholinguistic theory of interference and errors, and the theory of SLA, resulted in the rejection of its^[13] validity and expediency altogether by many linguists and language teachers (Fisiak 1980:10, 1981:6, Sajavaara 1977/81a), in many cases culminating in a total ostracism of the mother tongue from the FL classroom. In the words of Sanders (1976/81:21): “[CA] was pulled down from its pedestal as ungracefully and as unnecessarily as it had been hoisted up there following Linguistics across Cultures.” While the reaction rightly challenged the extreme claims made on behalf of CA, it wrongly overlooked aspects thereof which had not been invalidated but proven useful in FLT.^[14] Such misconstrued criticism of CL in general culminated with the 1968 Georgetown Roundtable Conference devoted to CA (Alatis 1968), considered a landmark in its development in the US in particular, and where it afterwards slowly died out as a result of the severe criticism (cf. Di Pietro 1976; although a re-evaluation of CA continued at the Pacific Conference on Contrastive Linguistics and Language Universals, held at the University of Hawai`i in 1971; Whitman & Jackson 1971).^[15] The conference introduced error analysis as a ‘contestant’^[16] to CA (Sajavaara 1981a:35) and a more powerful and accurate source of information about FLL difficulties (Banathy & Madarasz 1969). This soon became an autonomous field of study claimed to provide insights into SLA strategies.

As one consequence of empirical findings in the late 1960s and 1970s, transfer came to be seen as but one of several processed involved in SLA. Yet, it was still considered by some to be the most important one; e.g. Selinker (1972) saw it as the prime perpetrator of fossilisation, just as Nemser (1971) regarded it as one of the major determinants of the learner’s ‘approximative systems’. This led to investigations of the interaction of transfer with other factors and strategies.

The minimalist position

In the wake of the demise of the much-maligned CAH, the minimalist position emerged (as reflected in the work of e.g. Dulay & Burt 1972; the term not to be confused with Chomsky’s (1993, 1995) Minimalist Program brought up earlier in this dissertation), which sought to downplay the importance of the L₁ in favour of universal processes of language learning and processing (such as hypothesis-testing; Ellis 1994:310), emphasising the similarity between L1A and SLA. It dismissed evidence of the pervasiveness of interference in language contact situations as irrelevant to SLA, and also fell into the trap of contriving studies in a way which ensured few interference errors (eliminating structures with a potential for transfer and investigating ones where it is highly unlikely; Ellis 1994:311). At the same time, some researchers staking out a minimalist transfer position and only attending to learner errors ignored the positive contribution of the L₁ (Ellis 1994:304; in the 1980s Hakan Ringbom actually argued that positive transfer is much more influential than its villainous alter ego). Also, as Kellerman (1987) pointed out, it was forgotten that the absence of a structural feature in the L₁ may have as much impact on the L₂ as the presence of a different feature. He also made the point that the actual effect of the L₁ can only be established by counting error types rather than tokens—which was the methodology used in the minimalist position—as the number of the latter will depend on that of obligatory occasions for the use of a particular structure (e.g. article errors are likely to be more common than passive voice errors). In terms of word order studies, minimalist accounts also restricted themselves to attempts at explaining only apparent instances of L₁ influence. However, Odlin (1990) musters convincing evidence of transfer also where basic word order is concerned, dismissing e.g. alternative discourse strategy accounts as well as Rutherford’s (1983) and Zobl’s (1986) ‘universalist position’. He accounts for the relatively few instances of basic word order transfer in the literature by pointing out the relative lack of research on beginner learners and the high level of awareness that salient word order arouses, which may help learners monitor their production.

Resurgence of CA

There is now 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. …
In the face of increasing quantities of L2 data, researchers have begun to once again focus their attention
on language transfer, realizing that the baby had been mercilessly thrown out with the bathwater.
(Gass & Selinker 1992:7)

Providentially, despite its downfall transfer still remained of interest to SLA researchers, who reapplied CA as a tool for identifying potential areas of difficulty (Fisiak 1981a:7) and relocated transfer within a cognitive framework, focusing their endeavours on identifying the conditions inducing and inhibiting it (Ellis 1994:315). Actually, CA was never entirely on the defensive (cf. e.g. Schachter 1974; Wode 1978), as even the proponents of other approaches (EA or IL) incorporated it in their methodology, whether explicitly or implicitly (Sridhar 1980/81:210, 232). Sridhar (op. cit.:219) argues that CA cannot in fact be merely a subcomponent of EA because 1) a corpus-free theory of errors is needed, not only a taxonomic classification of data culled, and 2) ‘predictive’ CA brings to light areas of difficulty overlooked by EA (cf. the investigations in Dušková 1969a; Banathy & Madarasz 1969; Richards 1971b; Schachter 1974; Celce-Murcia 1978; inter alios). Corder (e.g. 1971) also pointed out that not all errors are directly observable. For instance, a learner’s utterance, though superficially well formed, may have been produced correctly ‘by chance’, by an algorithm different from that of the TL, i.e. owing to holophrastic learning, or with the actual meaning different from the intended one, or manifested in the systematic avoidance of problematic structures (cf. also Schachter 1974). Then, if anything, the turbulent controversy only seems to have contributed to a clarification and sobering reaffirmation of CA’s possibilities and limitations (Sridhar 1980/81:210). Most of the criticism levelled at CA in the 1960s was ably refuted in James (1971, 1980). CA can be defended and rationalised on the grounds that:
-          while not all of a learner’s problems are attributable to direct interference from the mother tongue, every experienced language teacher will confirm that a substantial number of persistent mistakes are due to the learner carrying over L₁ patterns into TL performance, and that the overall patterns of error do tend to be language-specific (Sridhar 1980/81:210; Swan & Smith 2001:xi);
-          if certain items are regularly substituted in the TL, this is with high likelihood engendered by L₁ interference, and what is called for is more CA, not less (Sanders 1976/81:23f.);
-          the very knowledge that a target item is nonexistent in the learners’ L₁ is useful in isolating a problem area, even if it can go no further (ibid.);
-          the non-occurrence of error does not necessarily invalidate the prediction itself; it may actually confirm it providing evidence for the avoidance of challenging structures by the learner (Corder 1973; Schachter 1974; Celce-Murica 1978);
-          the failure of predictions in particular instances (while they were borne out in scores of other empirical studies, e.g. Dušková 1969b; Bieritz 1974; Schachter 1974; inter alios) does not necessarily invalidate the theory, but calls for its refinement (Sridhar 1980/81:219). George (1972) estimated approximately ⅓ of all FL errors to be traceable to L₁ interference; hence, as long as transfer is one of the variables contributing to success or failure in FLL, CA should have a place in the methodology (Stockwell 1968);
-          CA is not incompatible with the generativist view of language learning as a process of hypothesis testing (Corder 1967), if one shifts the psychological basis of ‘interference’ from the behaviourist conditioning principle to something more akin to transfer of training, where the NL may be selected as one of the learner’s initial hypotheses (or ‘processing strategies;’ Sridhar 1980/81:220). In the words of Selinker (1992), “one can believe in language transfer without being a behaviourist. One wonders why this was not always clear;”
-          CA is an eminently useful instrument in materials design, able not only to predict areas of potential error—even if not pinpoint precise errors (Jackson 1981:204)—but also providing explanation and remedying many of those that actually crop up. Thus, it is able to provide an inventory of useful data for authors of textbooks and pedagogical grammars in at least a few areas (W. Marton 1972/81:165);
-          CA would come particularly helpful where the teacher does not have competence in the NL of the learners (nor the experience of having learnt a FL)—the basic knowledge necessary to understand and remedy interference errors (Jackson 1981:199);
-          CA may be valuable in overcoming learners’ pronunciation problems by identifying e.g. L₁ phonemes which will facilitate the acquisition of certain TL sounds, where such reference may be more effective than plain cross-linguistic contrast (Sanders 1976/81:23f.; for instance, Zybert (1997 & p.c.) points out that in Polish we have allophones isomorphic to some English main variants but with a different distribution, e.g. raised variant of [e] – front half-open unrounded [e^] in the palatal context, as in sieć or ziemia (vs. tense upper mid [e] English set, let, etc.), or central mid [Î] in bełkot; what is then needed is the learner first noticing them—since contrasts between non-discriminating phonemes in the L₁ are not perceived by the auditory system—and then developing a habit—fixed motor routine—where a known sound will be used in a new phonetic environment, “to extend prior knowledge to new phonetic contexts in such a way that new, automatized articulatory routines are developed;” op. cit.:103);
-          though a hierarchy of difficulties based on CA may not be of much use in syllabus sequencing, it is CA that helps determine the relative frequency and stylistic distribution of certain structures in both languages (Levenston 1971);
-          it can help to teach the learner to use idiomatic TL features in his/her speech, and thus stop sounding ‘foreign’ (Sanders 1976/81:30);
-          it can play a crucial role in both ‘didactic’ (selection, grading, and exposition of teaching items) and ‘methodic’ (actual classroom presentation) programming (Nickel & Wagner 1968);
-          the results of CA may provide useful criteria for the selection of testing items (Lado (1957); but consult Upshur (1962) for an opposing view);
-          and Lee (1968) points out that the critics of the lack of a 100% predictive ability forgot that the aim of CA was only to refer to “behavior that is likely to appear with greater than random frequency” (Lado 1968), never claiming that it accounts for all errors. James (1971:56) also pointed out that contrastive analysts had never claimed that L₁ interference was the sole source of error, adding prophetically that “it will probably turn out that many of the errors which are now not traceable to the L1, and are therefore attributed to L2 overgeneralization, will … be recognized as errors of interference” (for other voices agreeing that the claims of the EA of three decades ago were precipitate cf. e.g. Schachter & Celce-Murcia 1977; Jarvis & Odlin 2000; or Jarvis 2002 for positive and negative transfer involving articles)^[17].

Fortunately, despite the strong critical voices coming particularly from America and the fact that some of the confusion from the American scene was transferred to Europe, situation in the Old World—which adhered to its established contrastive tradition more faithfully—was entirely different (Fisiak 1981:5), largely unaffected by the turbulences. A sizeable number of both linguists and teachers in Europe and beyond still found CA useful in error analysis, materials design and language teaching (op. cit.:Preface). Apart from the Prague linguists, scholars across the continent produced numerous theoretical contributions in the two after-war decades (Orr 1953; Valtonen 1953; Glinz 1957; Kielski 1957-60; Krushel'nickaya 1961; Enkvist 1963; to name only a few), while applied CL found its proper perspective and place as well (Fisiak 1981:6). Several organised projects were launched at various centres of active research (Kawińska 1976; with the contrastivist movement very strong in Poland, including a project originated by Prof. Jacek Fisiak at the Adam Mickiewicz University in Poznań, resulting in a Polish-English contrastive grammar textbook based on the transformational framework – Fisiak et al. 1978; with a contrastive generative grammar by Krzeszowski (1974) also being an important landmark in the annals of contrastive theory formation; Sajavaara 1981a:37), with over a thousand papers and monographs written over that period (Sajavaara & Lehtonen 1975, 1981). Subsequent research (e.g. Odlin 1989; Selinker 1992; James 1998) reinforced the re-establishment of interlingual transfer as a—if not the—major factor in SLA/FLL, giving CL the green light de nouveau. The focus, obviously, had to depart from the original one, now moving towards reconciling the phenomenon with the cognitive, developmental perspective (Gass & Selinker 1992).

Concurrently, contrastive linguistic research gained momentum with the emergence and rapid development of corpus linguistics (allowing parallel compilation, cluster computing and complex queries) and advances in the automatic assignment of syntactic categories (PoS tagging), which are increasingly focusing on cross-linguistic studies, including massive bilingual corpora of authentic language material^[18], enabling more objective, reliable, high-quality empirical investigation and quantification of formerly mainly intuition-based judgments (Granger 2003:17f.; Rawoens 2006:3). Aijmer & Altenberg (1996:12) list the following advantages of bi-/multilingual corpora:
-          they offer new insights into the languages compared, likely to go unnoticed in studies of a solitary language and by no means obvious even to professionals;
-          they can increase our knowledge of language-specific and typological differences as well as universal features;
-          they illuminate differences between native and non-native texts as well as between source texts and translations;
-          they can find practical applications in e.g. language pedagogy, translation, and lexicography;
-          to which Granger (1996:45) adds the benefit of comparing e.g. learners’ output with that from a corpus of professional translations for determining rates of over/underrepresentation.

Among such multilingual corpora one can distinguish translation (parallel) corpora, containing source texts and their translations (unidirectional, if they contain originals in one language and their translations into another; bidirectional if they contain originals and translations in both; or multidirectional in the case of more than two languages), and comparable corpora containing original texts of the same genre or theme (e.g. juridical; Altenberg & Granger 2002; Granger 2003:20), the typology depicted in Figure 1:

Figure 1: Corpora in cross-linguistic research (Granger 2003:21)

Such corpora can be employed in two ways: as a source of underlying data on the basis of which hypotheses are formulated (the so-called ‘corpus-driven approach’), and as a tool for the verification of previously conjectured hypotheses (the ‘corpus-based approach’; Rawoens 2006:2).

Contrastive corpora and language teaching

Corpus-informed research has propelled the recent inductive FLT methodology of data-driven learning (DDL). This can be defined tout court as “the use in the classroom of computer-generated concordances to get students to explore regularities of patterning in the target language, and the development of activities and exercises based on concordance output” (Johns & King 1991:iii). In the FL classroom, aligned corpora are primarily employed in consciousness-raising exercises, where the students are presented with instances of language areas in the TL (alongside their aligned equivalents in their L₁) and asked to infer the underlying differences and/or similarities (as an example of the DDL approach Granger (2003:24) mentions the Online Chemnitz Internet Grammar^[19], which draws extensively on the English-German Translation Corpus). Though ideally suited for more advanced learners, there is evidence (St. John 2001) that it is also beneficial with beginner levels. Bilingual corpora have also been evinced to be an invaluable pedagogical resource in ESP classes (Kübler & Foucou 2003).

Plurilingualism and the role of contrastive language teaching in the CEF

Linguist – a talented fellow who has mastered the ability to make mistakes in more than one language.
—The LowComDom Online Dictionary^[20]

One of the key concepts in the Common European Framework of Reference for Languages (2001) is that of plurilingualism, closely tied with the aim of developing European citizenship, with an educated European able to get by in several languages. Standing in what we could term ‘political’ contrast to multilingualism, which denotes at least decent fluency in three or more languages, plurilingualism is content with—and promotes—incomplete linguistic competence in these.^[21] This seemingly undemanding objective is due to the Framework’s positive and pronounced emphasis on helping learners communicate with users of another language, “however laboriously and incompletely” (Morrow 2004:5). As such, it inextricably involves recognition of the role of already mastered languages:

those who have learnt one language also know a great deal about many other languages without necessarily realising that they do. The learning of further languages generally facilitates the activation of this knowledge and increases awareness of it, which is a factor to be taken into account rather than proceeding as if it did not exist. (CoE 2001:170)

In the process of language learning students ought to be made aware of the range of skills and awareness which they possess from learning and using their first^[22] and other languages. Since there are rarely ‘absolute’ FL beginners, as many rules, structures, expressions and lexemes transfer directly from their current knowledge base (where they may exist as common borrowings), the CEF promotes the view that most learners are not complete tabulae rasae, but already have some degree of competence in the TL (probably having had some contact with it in one form or another), making it the teacher’s duty to bring this resource to light and let it expand.

Another obvious corollary is that learners who have already had some experience of learning another language will have an advantage embarking upon successive ones, as they will be able to utilise the formerly acquired and well tried-and-tested skills and strategies. The more languages you know, the easier it is with successive ones. Consequently, more emphasis should be placed “on developing strategies and skills for ‘learning to learn languages’ since the learner can apply these skills to learning or acquiring other languages… Teachers sometimes assume that a beginner starts from scratch, but in fact most have experiences of other languages and skills and knowledge they can apply usefully to learning the new language” (Heyworth 2004:15).

Since one of the aims stated in the CEF is for the teacher to not only convey linguistic competence, but also—promoting learner autonomy and the development of reflective learning habits—train students to become successful language learners, “capable of acquiring the particular languages as they meet the need for them” (op. cit.:13ff.), this should immediately entail enabling them to utilise cross-linguistic comparisons.

A further rationale for a contrastive approach is connected with the CEF promoting intercultural competence, i.e. not just knowing what language/register is appropriate for use in particular circumstances (sociolinguistic competence), but also how this appropriacy differs between cultures (Heyworth 2004:14; see also the discussion in Chapter 8).

A contrastive-linguistics approach is also present under ‘mediation’ in the new division of skills^[23] replacing the traditional listening-speaking-reading-writing quadruple. Mediation basically involves being able to manipulate a text and the information contained within, thus such editing and modifying skills as summarising and elaboration, paraphrasing, reporting, simplification and register transfer, but also—and chiefly—translation and interpretation!

Round-up

Contrastive linguistics, once faded with time, is slowly earning renewed interest and recognition of its advantages. Such events as the conference on Multilingualism and Applied Comparative Linguistics (Brussels, Feb. 2006), and publications such as Learner English designed “to help teachers anticipate the characteristic difficulties of learners of English who speak particular mother tongues, and to understand how these difficulties arise” (Swan & Smith 2001:ix), or A Contrastive Approach to Problems with English (Willim & Mańczak-Wohlfeld 1997) provide evidence of the continuing vitality and importance of this field, and CA is still widely used even where this may not be overtly acknowledged. Nowadays, more emphasis than in the past is being paid to similarities rather than differences (e.g. Selinker 1992:172). The chief contribution of contrastive linguistics to the current ELT scene is probably that in establishing the ‘common core’ but, as it will become clear in Chapter 4, it also plays a role in the Language Interface Model.

sources: Paradowski, Michał B. (2007) Comparative linguistics and language pedagogy: concise history and rationale. In: Boers, Frank, Jeroen Darquennes, & Rita Temmerman (eds) Multilingualism and Applied Comparative Linguistics, Vol. 1: Comparative Considerations in Second and Foreign Language Instruction. Newcastle-upon-Tyne: Cambridge Scholars Press, ch. 1, 1–18, and (2007) Exploring the L₁/L₂ Interface. A Study of Polish Advanced EFL Learners. Institute of English Studies, University of Warsaw, pp. 34–49.

---

^[1] E.g. Pye & Kryszewska (2002, 2003) with grammar contrast boxes by Grzegorz Śpiewak.
^[2] Just as there is no universal language syllabus, there can be no generalisable cultural one: “[a] French learner of English needs a different syllabus and methods to a Greek, and different again from a Japanese.” (Byram 1997:4)
^[3] In this context by the L₁ I will understand, in line with Atkinson (1993:3), the common language shared by the students, though not necessarily the mother tongue of each of them (taking as an example Catalan or Basque in Madrid).
^[4] Although I do not completely agree with de Echano’s (1977) proposal that truly effective simplified readers should be made convergent with the NL of the target reader population.
^[5] These insights will be complemented with more arguments in the Supplement G.
^[6] Who at the same time corresponded on Persian calligraphy with Prince Adam Jerzy Czartoryski (Nov. 1786, National Museum in Cracow).
^[7] Even if the concept were defendable, it would be worthwhile considering whether some of the more difficult constructions had not more efficiently be taught first, consequently activating more levels of the ‘accessibility hierarchy’ simultaneously. Gass (1982) and Eckman et al. (1988) demonstrated that learners who have successfully been taught a typologically marked structure will generalise their learning to less marked ones as well, which implies enhanced efficiency of the method.
^[8] American linguists did also contribute to more theoretically oriented CS in the areas of bilingualism and language contact phenomena (e.g. Weinreich 1953; Haugen 1953, 1954, 1958), although to a much lesser extent (Sajavaara 1977:10).
^[9] Psycholinguists revealed a ‘spectacular asymmetry’ between linguistically defined complexity of derivation and psychological difficulty (James 1979/81:61).
^[10] The L₂-er may e.g. make the L₁ the matrix for reference in the acquisition and use of TL elements and relations (a ‘compound setting’), or try and keep the languages as separate as possible (a ‘coordinate setting’; W. Marton 1972/81:159f.).
^[11] An account of a language is descriptively adequate if it provides a complete and explicit description of the rules and items comprising NS competence. It satisfies the criterion of explanatory adequacy if it also provides an explanation for how this competence is achieved in L1A, SLA, or FLL.
^[12] Somewhat ironically, CA—whose downfall was partly due to Chomsky’s criticism of the underlying assumptions—gained fresh impetus by his revolution, both enabling comparisons and contrasts to be explicit, precise, and insightful to a degree undreamt of two decades before, and—by positing the existence of universals—providing an apparently more solid theoretical foundation (Sridhar 1980/81:209, 214; but cf. Bouton 1975). Thus, owing to the application of generative theory, CA only gained in sophistication and rigour (Krzeszowski 1971). Sridhar (1980/81:213) itemizes three aspects of the transformational framework which profoundly inspired contrastive studies:
-          the universal base hypothesis, which— weighed against the structuralist relativity hypothesis—provided a sounder theoretical foundation and a base for comparability, facilitating the formulation of similarities and differences in a uniform manner;
-          the distinction of two levels of linguistic organisation (deep vs. surface structure), enabling capturing and representing the intuitions of bilinguals about the translational equivalence of superficially disparate utterances;
-          the adaptation of mathematical models and notations making descriptions and comparisons of linguistic phenomena precise and explicit.
^[13] Incidentally, note how the acronym ‘CA’ has since symptomatically shifted its meaning to denote Conversation Analysis, while PSiCL from the original Papers and Studies in Contrastive Linguistics has metamorphosed into Poznań Studies in Contemporary Linguistics!
^[14] But see Sajavaara (1981a) for a recognition of this crisis in the theoretical branch only.
^[15] But see Sridhar’s (1980/81:209) evaluation of the papers from this and two more conferences (held in Cambridge – Nickel 1971a, and Stuttgart – Nickel 1971b) presenting the scholars as generally sanguine about CA’s potential.
^[16] W. Marton (1972/81:162, 165) points out that EA and CA are not an exclusive alternative, but complementary phenomena, with EA only pointing out the most frequent mistakes/errors and their areas of occurrence, but not providing any explicit explanation. Similarly, Sridhar (1980/81:221-3) argues that typical EA—with the goal of designing pedagogical materials and strategies of correcting and eradicating learners’ errors—went little beyond “impressionistic collections of ‘common’ errors and their taxonomic classification;” making little attempt to define error, or to systematically account for its occurrence. The underlying assumption was that, by identifying areas of difficulty, EA could help in (ibid.):
-          sequencing the presentation of material, with the more difficult items following the easier ones;
-          determining the relative degree of emphasis, explanation, and practice required to put the items across;
-          devising remedial lessons and exercises;
-          selecting items to test proficiency.
Sridhar thus saw the role of EA is as a ‘testing ground’ for the predictions of CA as well as to supplement them.
^[17] I am grateful to Prof. Terence Odlin for bringing these sources to my attention.
^[18] E.g. the English-French Hansard Corpus available from the Linguistic Data Consortium at http://www.ldc.upenn.edu/Catalog/LDC95T20.html
^[19] http://www.tu-chemnitz.de/phil/InternetGrammar/
^[20] Retrieved from http://www.lowcomdom.com/dictionary/l.html
^[21] The European Commission uses the term in a slightly different sense than the Council of Europe; that of promoting cultural diversity across the EU, with a person deemed plurilingual when able to communicate in the 2+ languages (Leonard Orban, EU Commissioner for Multilingualism, p.c., 7 Mar 2008).
^[22] Annually on February 21, UNESCO observes International Mother Language Day to “preserve linguistic diversity and promote multilingual education.” In 2007, the theme was—significantly—the linkages between the mother tongue and multilingualism.
^[23] The remaining three being ‘interaction’ (combining listening and speaking or reading and writing in a coherent exchange), ‘reception’ (listening and reading comprehension), and ‘production’.