0.1 The present study: purpose, source of data, statistical results

This study was undertaken for two main reasons: the need for a general survey of grammatical errors made by advanced German learners of English, and the need to look more carefully at the difficult question of explaining why some of these errors are errors--why they are wrong in terms of the rules of English grammar. The second problem is one that every teacher of English, including those who happen to be native speakers of English, will recognize. It is also one that becomes progressively more complex as the learners' level of competence increases. Students at the Oberstufe or university level not only make mistakes that are on the whole more difficult to explain than those of less advanced learners, but as adults they also require a more cognitive therapeutic approach than one involving mere correction and practice. The adult learner, in many cases, wants to know why his error is wrong and how he can avoid it in the future. In other words, he wants grammatical rules. The rules one finds in the standard grammars, however, are not necessarily optimal, from either a theoretical or a pedagogical point of view, and in some cases, learners' errors require explanations or rules that do not appear at all in existing grammars.

This is an error grammar--a partial grammar or collection of rules designed to explain why certain errors are wrong. The rules presented here are intended to explain some of the most common grammatical errors made by advanced German learners of English. "Grammatical" is a vague term, but without entering the morass of theoretical debate as to which aspects of language are syntactic, lexical, or semantic, most English teachers will recognize the errors discussed here as grammatical, i.e. as opposed to phonetic or lexical (vocabulary) errors. This is also a kind of pedagogical grammar, in that it is intended to help a language-specific group of learners (Germans) learn English, but there are very few contrastive statements about English and German grammatical structures, and no explicit pedagogical exercises. (For a recent discussion of the pedagogical implications of error analysis and contrastive linguistics cf. Nickel 1980.) Some of the examples might be used directly as pedagogical exercises; others, extracted from their original context and varying a great deal in any case in their degree of (unacceptability, are far too problematic for this purpose.

The construction of practice exercises, though, is a problem in itself, and not the subject of this work. The purpose here is to discuss the grammatical problems presented by the analysis of various specific errors. Explicit cognitive explanation of grammatical structures, we now realize, can in fact play an integral role in language teaching and learning, just as it does in other fields of human learning (cf. Diller 1971 for a review of the arguments). We know a lot more today about language and grammar (especially English grammar) than we did twenty years ago, due in large measure to the impetus given to linguistic research by transformational generative theory. The problem now is that linguistic argumentation has reached such a rarefied level that very few of whatever insights may be found there are readily accessible or ever find their way to language teaching professionals, not to mention language learners. The attitude represented in this treatise is that the current gap between theory and practice in linguistics is unnecessarily wide, and that useful discussion of grammatical topics can take place without being couched in a complicated, abstract theoretical apparatus, but at the same time without over-simplifying the issues.

Since this study is data-based' and data corpora are necessarily limited, it cannot be exhaustive, and important error types have undoubtedly been omitted. Nevertheless, the range of errors dealt with is broad enough to be considered representative of the most common grammatical errors made by advanced German learners, and thus should be of interest to curriculum planners and textbook writers, as well as to teachers. The approach to errors adopted here is quite different from that found in most of the literature in the field of error analysis. (For a review of the literature, cf. Schachter 1977 and Palmberg 1980.) First of all, "explaining" errors in most cases (Burt and Kiparsky 1972 is an exception; cf. also Swan 1980, Bischoff 1974) has meant explaining not why the error is an error, but why the learner made the error. It only seems logical, though, that a purely linguistic analysis should precede psycholinguistic investigations of the possible causes of errors, such as interference from the source language or overgeneralization of rules in the target language. In what sense can we talk about overgeneralization of rules, for example, unless we have an adequate description of those rules to begin with? Furthermore, one possible causal factor of errors is surely the degree of complexity of the linguistic structures (in the target language) involved, and the only way to determine just how complex these structures are is to try to formulate descriptive rules to account for them.

There has also been a considerable amount of discussion about the difference between "errors" and "mistakes" and about the significance of the frequency of occurrence of errors (cf. Corder 1967, 1971, 1972, Zydatiß 1976, 1977). These are, of course, interesting and important questions, but they are not taken into consideration here for practical reasons. First of all, the collection and organization of- data is enormously more complicated if these distinctions are made. Secondly, in most situations a learner's performance in the foreign language is evaluated simply by the number of errors he makes, whether they seem "elementary" or not (i.e. something he should know), and the degree to which they hinder communication. The question of whether the error is "systematic" or not (i.e. an error as opposed to merely a mistake) cannot be considered in most situations, since to answer it would require a good deal more investigation than is normally feasible. (Does the mistake occur regularly? Can the learner avoid the mistake when not under stress? Is the mistake self-correctable--i.e. can the learner correct it himself if it is called to his attention?) Similarly, in most situations it is quite impractical to consider the frequency of occurrence of a particular error in relation to the frequency of occurrence of the corresponding correct structure (a relationship which would presumably reflect the degree of learning difficulty of the structure involved). From a realistic point of view, the learner's performance in the foreign language is evaluated according to the errors he makes without regard for these finer distinctions.

The data for this study was collected over a period of four years from a variety of written and oral sources: Sprachklausuren (four hour free compositions), in-class essays, taped monologues and class discussions. The subjects were German university students, all of whom had had nine years of English in secondary school and were preparing to become English teachers themselves. Approximately 1200 sentences containing errors were extracted from 300 separate text samples representing 100 students (male and female, in various semesters). All of the sentences occurred in free writing or speech; texts resulting from translation, Nacherzählungen and other kinds of more structured exercises were not considered. The only guidance involved in the Sprachklausuren, in-class essays and taped monologues (recorded by the students at home) was in the form of a sentence or short paragraph presenting a topic to which the student was supposed to respond. All of the examples were taken from a context in which the grammatical structure concerned seemed unacceptable or at least odd, although out of context many of the sentences may seem acceptable, or a context might be constructed in which they would be acceptable. Some of the examples, of course, are more or less acceptable than others.

The errors are classified, where possible, according to the grammatical rule which they violate. This method avoids some of the problems involved in error classification (cf. discussion below in 0.2), though some degree of arbitrariness is inevitable. More importantly, this classification allows immediate focus on the aspect of the error of primary concern, namely, the grammatical rule which explains why the error is wrong. In the detailed table of contents the number of errors that occurred in each category and subcategory is expressed as a percentage of the total. This percentage figure represents the relative frequency of occurrence of the various categories of grammatical errors (rounded off to the nearest whole number). Errors accounting for less than .5 percent of the total are marked with a plus sign (+). The figures for the general categories are as follows:

Within the noun phrase the percentage of errors in various subcategories breaks down approximately as follows:

Within the verb phrase the percentage of errors in various subcategories is:

This sort of general statistical picture is helpful in curriculum and textbook planning, since it gives an overview of the areas in which errors are to be expected and where teaching emphasis should be placed to combat them.

A more detailed picture can be gained by considering-the percentage of errors occurring as violations of individual rules (see detailed table of contents). These are the smallest categories, though they do not in all cases correspond to rules. The category or section with the largest percentage of errors, for example, covers the misuse and omission of prepositions (4.2) and there is obviously no single rule that can adequately describe all of these errors. (Of course, if one were to expand the number of rules indefinitely, the lexical rules specifying the usage of the individual prepositions--as in a dictionary--could be listed, but this is not practicable here.) Similarly, it was not felt worthwhile to formulate rules for the use of the modal verbs (2.6.1-6), the morphology of individual verbs (past tense and past participle--cf. 2.9), or the use and position of adverbs (8.4-5). Of the 104 rules (or sections, if there is no rule), 49 account for 1% or more (each) of all the errors; 55 rules account for a very small number of errors--less than .5% each. There are only 16 rules/sections that account for 2% or more each of the-total number of errors. These are:

Thus we see that by far the greatest number of errors involve the use of prepositions. The second most frequent error (1.1) involves generic reference--usually the use of the definite article where it is not called for. 1.2.1 is for the most part the reverse error--the failure to use the definite article where it is required. 1.4.1 involves the failure to distinguish between count and mass nouns--more specifically, the misuse of the plural and the indefinite article with mass nouns. 2.4.1 and 2.4.2 both involve perfective aspect, i.e. the use of the simple present or past where the present perfect is required, or vice versa. And so on. This breakdown of the relative frequency of errors in categories that, for the most part, correspond to rule violations allows us to point with a fair degree of precision to the rules or structures of English grammar which present the most difficulty to advanced students--at least those which occasion the greatest number of errors.