The heyday of EA
The heyday of EA was short-lived, however. In 1971 Corder had published a paper entitled ′Idiosyncratic dialects and error analysis′.The concept of idiosyncratic dialect (ID) is a development of the 1967 concept of ′transitional competence′. It also has strong affin- ities to Selinker′s ′Interlanguage′ (1972) and an uncanny resemb- lance to Nemser′s ′approximative system′ (1971). For Corder, an ID is ′a special sort of dialect′, special in that while some of its rules are held in common with speakers of other dialects of the TL (or target dialect), these are too few to ensure interpretability of the learners′ utterances by others. There are also too few indi- viduals sharing the rules of anyone ID to allow us to say that it has a community of speakers. In other words, it is not a social dialect but an idiosyncratic one. Some of these claims have gone unchal- lenged. It is, for example, demonstrable that learners of any given FL who share the same L1 and have been taught under similar conditions with the same texts and syllabus do emerge speaking the same ′social′ dialect of that FL. They encounter few problems understanding each other′s utterances rendered in that dialect, but outsiders, including speakers of the target dialect, that is ′native speakers′, find it unintelligible (see Lane, 1963). Corder, like Nemser, proceeds to argue that there are other sorts of ID besides those of FL learners: the language of poets and poetry, aphasic language, child language. But he insists that there is a difference, and this is where EA goes to the gallows. It is legit- imate, Corder contends, to compare a poet′s or an aphasic′s ID to ′normal′ dialect of a prose writer and a neurologically healthy person, using an ′approach that is essentially that of ″error analysis″, a type of bilingual comparison′ (Corder, 1971: 150, my emphasis). It is not deemed legitimate, on the other hand, to compare the child′s or the FL learner′s ID to the dialect of adults or of native speakers of the FL respectively. [unique_solution]The reason, Corder argues, is that the poet and the aphasic are – or once were – speakers of the standard dialect, and their deviance from the standard deselVes the label because it is deliberate or pathological respectively. On the other hand, the child or the FL learner are neither deliber- ately nor pathologically deviant in their language, so it would be wrong to refer to their repertoires as erroneous. In the case of the FL learner, use of the label ′error′ would be particularly inappro- priate ′because it implies wilful or inadvertent breach of rules which, in some sense, ought to be known′ (ibid.: 152). Reference to ′error′ would only be justifiable if the rules had been deliber- ately flaunted or caused by performance factors, that is, if they were ′cases of failure . .. to follow a known rule′ (ibid.: 152). Corder uses two further arguments against calling the FL learner′s ID ′erroneous′, ′deviant′ or ′ill-formed′. First, he objects that to do so is ′to prejudge the explanation of the idiosyncrasy′ (ibid.: 152). We study an ID or IL in order ′to discover why it is as it is … to explain it′ (ibid.: 153). However, as we shall presently see, it is not the ambition of EA to jump to conclusions in the way Corder implies. The explanation (or ′diagnosis′) of a unit of learner language that does not match its equivalent in the TL is in no way prejudged by the simple act of calling it an error. Corder′s second argument is aimed at those who call a learner′s sentences ′ungrammatical′: this is wrong, he insists, since ′they are in fact grammatical in terms of the learner′s language′ (ibid.: 153), that is in terms of the learner′s IL or ID grammar. This is uncontentious: when we say that something is ′ungrammatical′ we assume that this means ungrammatical in terms of the grammar of the lan- guage which the learners claim or believe themselves to have been speaking or writing at the time. Certainly everything is true ′unto itself′, but this is a truism that won′t take the learner very far. An analogy might help clarify the point. The mafia is said to have its own rules of conduct. We see it in operation in a novel like The Godfather. These rules are perfectly acceptable to the mafia, of course, and if this were not the case, it would amend them. But they are not acceptable to society at large, beyond the mafia: they are seen to be ′wrong′, in fact criminally and wilfully wrong, and those who use them are required to mend their ways, or face dire penalties. A telling difference between mafiosi and FL learners is that the former do not want their behaviour to conform to the standard, whereas most FL learners do. For as long as FL learners are prepared to see and call themselves learners, the assumption that they wish to conform is surely a reasonable one to make. Other, more direct, criticism of EA followed Corder′s. Bell refers to EA as ′a recent pseudoprocedure in applied linguistics′ (1974: 35), and attacks EA for its poor statistical inference, the subjectiv- ity of its interpretations of errors, and its lack of any predictive power (something any scientific procedure must have). There was more emphasis on such statistical and objective rigour when Bell wrote than there is now, and we have become more realistic and modest in our demands of a science. Hammarberg points to the ′insufficiency of error analysis′, which for him lies in its one-sided practice of ′analyzing out the errors and neglecting the careful desсrіption of the non-errors′ (Hammarberg, 1974: 185). This is short-sighted, he claims, because it conceals from teachers exactly the information they could put to good use, information about potential errors that the learners somehow manage to avoid com- mitting. Knowing how learners avoid certain likely errors is the first step to discovering how to help the same learners avoid the errors they fail successfully to avoid. This is a valid point of course, and is related to research (Cohen, forthcoming) that seeks to identify the strategies used by good FL learners so that these same strategies can be taught to not-so-good learners. Schachter (1974) also discovered what she saw as a fundamental flaw in EA – a fail- ure to recognize that learners have a tendency to avoid TL items they are not sure about, and so not to commit errors which they would be expected to commit. By 1977 Schachter and Celce-Murcia were entertaining a number of reservations about EA. We return presently to the strategy of avoidance and those other reservations. It seems, then, that EA was earmarked for obsolescence by the mid-1970s, and the theoretical ground was being cleared for the new IL paradigm. However, work in EA did not come to a sudden halt – as the rest of this book testifies. People continued to do EA despite the odds. Some fought a rear-guard action: ZydatiB (1974) offered a ′kiss of life′ for EA. Rocha (1980) suggested that ′objec- tivity′ can be salvaged in EA by submitting putative errors to panels ofjudges. Abbott (1980) answered some of Bell′s worries by sug- gesting ways to increase the rigour of EA. F£erch (1978), as if responding to Hammarberg (1974), offered Perfonnance Analysis as an alternative to EA that does take into account both errors and non-errors. Others were content to work in a theoretical vacuum, for the sake of doing something of practical value: I refer here to work on error gravities (James, 1974, 1977; Johansson, 1978; Hughes and Lascaratou, 1982; Davies, 1983; McCretton and Rider, 1993). Norrish (1983) produced an excellent account of EA for teachers. Others played safe and aligned themselves on the new IL para- digm, and, encouraged by the parallel communicative movement in FL teaching methodology, argued that errors and their ′creat- ive′ authors have to be treated with due respect (Edge, 1989; Page, 1990). It has also been a period for publishing collected papers on EA: Richards (1974), Svartvik (1973), Page, Harding and Rowell (1980). It is my contention that EA has become a more wide- spread practice than it is given credit for. The main objective of the next chapter will be to showjust how wide the scope of EA is.