English Inversion in Second Language Acquisition of German Native Speakers

Table of contents

ABSTRACT

LIST OF FIGURES

LIST OF TABLES

ABBREVIATIONS AND SYMBOLS

1 INTRODUCTION
1.1 Preamble
1.2 Review of related studies
1.3 Explicit Hypotheses and scope of the study

2 SECOND LANGUAGE ACQUISITION
2.1 History and theories of SLA
2.1.1 The Contrastive Analysis Hypothesis
2.1.2 Error Analysis
2.1.3 The Interlanguage Hypothesis
2.1.4 The Monitor Model
2.2 Reappraising transfer

3 TOWARDS AN UNDERSTANDING OF INVERSION
3.1 Inversion types in English
3.2 Theoretical preliminaries
3.3 Inversion in English and in German
3.4 Summary

4 METHOD
4.1 Participants
4.1.1 Learner-based factors
4.1.2 Language-based factors
4.2 Material
4.2.1 Questionnaire 1
4.2.2 Questionnaire 2
4.2.3 Questionnaire 3
4.3 Procedure
4.4 Coding issues

5 ANALYSIS AND DISCUSSION OF THE RESULTS
5.1 Learner judgements of negative inversion
5.2 Learner judgements of locative inversion with different semantic values of its initial PP
5.3 Learner judgements of locative inversion with varying degrees of subject complexity
5.4 Discussion

6 CONCLUSION

REFERENCES

APPENDIX

DEUTSCHE ZUSAMMENFASSUNG DER ARBEIT

Abstract

On the basis of questionnaire data, this study seeks to find a tendency for native language influence in the judgement of German-speaking learners of English. Seeing that the concept of language transfer has received diverse weight in the history of second language acquisition (SLA), the present account illuminates the issue anew, with a specific linguistic concern. Inversion structures, though not canonical, are very frequent in the German language, whereas the English language offers comparatively rare environments which trigger or allow for inversions. As previous studies found, language transfer, in regard to word order, emerged when the native language exhibited flexible word order, and the language to be learnt, in contrast, had a rather fixed one. This gave rise to the assumption that German natives would generally accept subject-verb inversion in declarative sentences in English, even if for a native speaker of English the structure would not be acceptable. Thus, the second language learners are expected to score high on the acceptability rate for the majority of instances of inversion in English, since the German equivalents of the test items are basically quite acceptable in German.

List of figures

Figure 1 Raw frequency; neginv 1–5 (Q1), rated by course 1 (13 subjects)

Figure 2 Raw frequency; neginv 1–5 (Q1), rated by course 3 (23 subjects)

Figure 3 Raw frequency; negisv 6–10 (Q1), rated by course 1 (13 subjects)

Figure 4 Raw frequency; negisv 6–10 (Q1), rated by course 3 (23 subjects)

Figure 5 Raw frequency; negipa 11–15 (Q1), rated by course 1 (13 subjects)

Figure 6 Raw frequency; negipa 11–15 (Q1), rated by course 3 (23 subjects)

Figure 7 Raw frequency; pplocs 1–5 (Q2), rated by course 2 (24 subjects)

Figure 8 Raw frequency; pplocs 1–5 (Q2), rated by course 7 (22 subjects)

Figure 9 Raw frequency; pploct 6–10 (Q2), rated by course 2 (24 subjects)

Figure 10 Raw frequency; pploct 6–10 (Q2), rated by course 7 (22 subjects)

Figure 11 Raw frequency; pplocc 11–15 (Q2), rated by course 2 (24 subjects)

Figure 12 Raw frequency; pplocc 11–15 (Q2), rated by course 7 (22 subjects)

Figure 13 Raw frequency; sloci 1–5 (Q3), rated by course 4 (24 subjects)

Figure 14 Raw frequency; sloci 1–5 (Q3), rated by course 5 (16 subjects)

Figure 15 Raw frequency; sloci 1–5 (Q3), rated by course 6 (22 subjects)

Figure 16 Raw frequency; slocl 6–10 (Q3), rated by course 4 (24 subjects)

Figure 17 Raw frequency; slocl 6–10 (Q3), rated by course 5 (16 subjects)

Figure 18 Raw frequency; slocl 6–10 (Q3), rated by course 6 (22 subjects)

Figure 19 Raw frequency; slocs 11–15 (Q3), rated by course 4 (24 subjects)

Figure 20 Raw frequency; slocs 11–15 (Q3), rated by course 5 (16 subjects)

Figure 21 Raw frequency; slocs 11–15 (Q3), rated by course 6 (22 subjects)

List of tables

Table 1 Questionnaire 1 – Total of test items

Table 2 Questionnaire 2 – Total of test items

Table 3 Questionnaire 3 – Total of test items

Table 4 Possible scores for all test items

Table 5 Overview of the distribution of the questionnaires

Table 6 Raw and relative frequency data (number of responses) – Questionnaire 1 (negative inversion), combined results of course 1 and 3 (with a total of 36 subjects)

Table 7 Raw and relative frequency data (number of responses) – 51 Questionnaire 2 (locative inversion, initial PP), combined results of course 2 and 7 (with a total of 46 subjects)

Table 8 Raw and relative frequency data (number of responses) – Questionnaire 3 (locative inversion, initial PP), combined results of course 4, 5, and 6 (with a total of 62 subjects)

Table 9 German equivalents of all test items of Q1 (NegInv)

Table 10 German equivalents of all test items of Q2 (LocInv, initial PP)

Table 11 German equivalents of all test items of Q3 (LocInv, subject complexity)

Table 12 Raw and relative frequency data (number of responses) – Questionnaire 1 (negative inversion), filled in by course 1 (with a total of 13 subjects)

Table 13 Raw and relative frequency data (number of responses) – Questionnaire 1 (negative inversion), filled in by course 3 (with a total of 23 subjects)

Table 14 Raw and relative frequency data (number of responses) – Questionnaire 1 (negative inversion), filled in by course 1 and 3 (combined, with a total of 36 subjects)

Table 15 Raw and relative frequency data (number of responses) – Questionnaire 2 (locative inversion; semantics of PP), filled in by course 2 (with a total of 25 subjects)

Table 16 Raw and relative frequency data (number of responses) – Questionnaire 2 (locative inversion; semantics of PP), filled in by course 7 (with a total of 22 subjects)

Table 17 Raw and relative frequency data (number of responses) – Questionnaire 2 (locative inversion; semantics of PP) filled in by course 2 and 7 (combined, with a total of 46 subjects)

Table 18 Raw and relative frequency data (number of responses) – Questionnaire 3 (locative inversion; subject complexity), filled in by course 4 (with a total of 24 subjects)

Table 19 Raw and relative frequency data (number of responses) – Questionnaire 3 (locative inversion; subject complexity), filled in by course 5 (with a total of 16 students)

Table 20 Raw and relative frequency data (number of responses) – Questionnaire 3 (locative inversion; subject complexity), filled in by course 6 (with a total of 22 subjects)

Table 21 Raw and relative frequency data (number of responses) – Questionnaire 3 (locative inversion; subject complexity), filled in by course 4, 5, and 6 (combined, with a total of 62 subjects)

illustration not visible in this excerpt

Abbreviations and symbols

illustration not visible in this excerpt

1 Introduction

1.1 Preamble

Anyone who has ever learned a foreign language is familiar with a range of uncertainties when it comes to situations in which actual confrontation with this language is required. A frequently observed (often in language teaching environments) habit of insecure language learners is to fall back on their native language by transferring (phonological, syntactical, etc.) knowledge to the language which is learnt. Although the influence of the mother tongue¹ in second language acquisition has to be seen as only one amongst a variety of factors which have an impact on the learning process, it is an effective strategy for learners to compensate, for instance, communicative gaps (which are caused by a lack of second language knowledge). In teaching, for instance, knowledge of specific transfer phenomena can help to explicitly detect major sources of error.

In their practical guide for teachers of English, Swan and Smith (1988) explicitly point out which errors are typical of German learners, who are learning English as a foreign language, and which are said to be directly related to native language influence. Particularly illustrative for the central purpose of this study is the following example demonstrating an error in word order:

If the subject of a main clause is preceded by anything other than a conjunction, the subject and verb are inverted [in German, T.S.]:

*On Tuesday have we a holiday. [emphasis in the original]

(Swan and Smith 1988:36)

The grammatical structure exhibited here (erroneously!) is called inversion of subject and verb - although in German inversion principally represents a deviance from the basic word order, namely SVO (though this is not completely accepted), it is much more productive than in English, a language which is more fixed in terms of word order. Inversion per se is highly complex phenomenon, and numerous attempts have been made (at least in English) to elucidate its purpose.

Primarily, the interest of this study is how learners of English (as their first foreign language) with German as their native language background evaluate inversion in English declarative sentences. To what extent will different types of inversion structures under diverse conditions be accepted? And what is more, how can cross-linguistic influence account for such observations? In order to attain reliable insights, three questionnaires have been conducted which were submitted to six groups of German-speaking learners of English.

The paper is structured as follows. In the remainder of this section, some related word-order studies are briefly discussed; subsequently the underlying hypotheses of this study are explicitly pointed out. Section 2 gives a general introduction to the study of second language acquisition (henceforth SLA) by referring to specific keywords, literature, and other important issues in the field, and by illuminating (chronologically) major theories on which the contemporary discipline is based. As it is of central interest in this study, a separate account on the concept of transfer follows. Section 3 provides a comprehensive theoretical fundament in order for the reader to grasp the complex phenomenon of inversion, and highlights relevant issues regarding this study. Section 4 contains an extensive description of methodological issues applied in the survey. An analysis and discussion of the results can be found in section 5. Some implications for further research are finally discussed in section 6.

1.2 Review of related studies

Even if word order phenomena have been subject to a number of recent studies in SLA, inversion in declarative sentences in relation to issues of native language influence has not been investigated as a subject of study in its own right, as it is in the present paper - at least not with regard to German learners. Furthermore, to my knowledge, no special consideration has been devoted to the investigation of the particular inversion types in the context of SLA which were to be judged in the survey that provides the spine of this study.

Nevertheless, there are several studies which discovered that learners whose native language exhibits flexible word order tend to transfer these word order patterns into their learner language system, which then causes syntactic errors in the target language output if this is a language possessing rigid word order (cf. Selinker 1969, Bates and MacWhinney 1981, Gilsan 1985). These findings can be seen as a general baseline for the predictions made in this thesis. Other findings, though, revealed that beginners in language learning (whose native language exhibits flexible word order) tend to cling to a fixed word order in the second language in order to rely on at least one constant in their state of initial insecurity. This observation by Pienemann (1981) was based on interview data from three Italian girls who were acquiring German as a second language, and might serve as an explanation of the sometimes unexpected results of this study. Furthermore, Ilomaki (2005) investigated a number of phenomena regarding cross-linguistic influence by accounting for errors; for what she conducted an error-recognition and a translation test with Finnish-speaking and English- speaking learners of German. An analysis of the errors revealed that the English learners in particular (Finnish has flexible word order) were prone to avoid the inversion of subject and verb when German necessarily demanded such a structure. This may support the assumption that in the reverse case (German learners of English) as regards word order, difficulties in terms of cross-linguistic influence might also be possible, as English and German exhibit a different degree of markedness concerning non-SVO patterns (cf. section 3.3), such as inversion.

As concerns the investigation of subject complexity in locative inversion, which will be one part of the general focus of my survey, the corpus-study of Lozano and Mendikoetxea (2006) should be mentioned. Their research interest dealt with the written production of post-verbal subjects under specific conditions of Spanish and Italian learners of English as a second language. Amongst other predictions, they hypothesised that the tendency for the occurrence of subject inversion increases when “the subject is long and ‘heavy’” (2006:1). This claim was unambiguously confirmed by the result of their examination. Thus, it can be stated that learners (at least from Romanic language background) typically tend to invert subjects which are ‘heavy’², in contrast to ‘light’ subjects, which occur in pre-verbal position. This topic will be discussed further in section 4.2.3.

1.3 Explicit Hypotheses and scope of the study

Drawing on previous work in SLA (which will be outlined in detail below), but originally inspired by observations gained from my own foreign language teaching experiences, this survey generally seeks to find evidence of native language influence in the judgements of German-speaking learners of English. The investigation is restricted to SLA in a classroom setting. The data was obtained by asking the learners - on a perceptive level - for there judgement of a list of English sentences by rating them in terms of acceptability. The groups being observed in the survey are not representative samples - rather, this study attempts to reflect a tangible tendency either in favour of or against the following predictions. The central research questions are:

-Does native language transfer have an impact on the judgement of English inversion structures by German-speaking learners?
-Do different degrees of markedness for one phenomenon in two languages entail language transfer?
-How can the issue of instruction be taken into account in this context?

These can be specified in formulating the following predictions (which are further specified in sections 4.2.1-4.2.3):

H1 Due to native language transfer, inversion structures which are acceptable in English are predicted to be rated highest on the acceptability scale by the learners.

H2 Generally on the same basis as H1, inversion structures which are least acceptable or unacceptable in English are not expected to be completely rejected, rather to achieve also a considerable rate of acceptability; if so, those cases can be categorised as resulting from interference (negative transfer).

H3 Inversion structures involving a different degree of subject complexity are predicted to score highest for the most complex (long) subjects; the acceptability rate will gradually decrease down the scale via intermediate subject complexity towards short subjects.

2 Second Language Acquisition

Serious social, political, and industrial developments of modern times were accompanied by an increasing interest in the nature of language learning and the subsequent intense study thereof. This curiosity set in around the middle of the twentieth century³. With the incipient globalisation of the world, the growing cross-cultural interaction entailed an exigent need to facilitate communication. This, in turn, called for the profound investigation of how languages are learnt, and how they most effectively have to be taught, as well as the implications linked to these questions.

There is a relatively concise number of introductory literature and handbooks on second language acquisition which give a general overview of the discipline (cf. Klein 1986; Ellis 1997a, [and rather comprehensive] 1997c; Gass and Selinker 2001; Mitchell and Myles 2004; Lightbown and Spada 2006; Saville-Troike 2006; as well as the handbooks edited by Ritchie and Bhatia 1996; Bausch, Christ, and Krumm 2003 [particularly focusing on issues concerning instruction and SLA]; Doughty and Long 2005; inter alia), as well as works especially addressing questions concerning SLA research (cf. Ritchie [ed.] 1978; Larsen- Freeman and Long 1994; Ellis 1997b; Byrnes [ed.] 1998; Aguado [ed.] 2000; Gass and Mackey 2005; inter alia). In the following, I will adopt the general tradition pursued by the majority of these works; that is to initially clarify the sometimes misleading and inflationary used terminology in the field.

Second language acquisition is the term used for the field of science originating in the US, which is rooted in first language acquisition (FLA) research (cf. Edmondson and House 2000:15). Since the work of Krashen (1978, 1982, inter alia) it is sometimes distinguished between the two notions learning and acquisition of languages, whereas the former implies that learners experience a language in a conscious process, which is in a way controlled and explicit (by instruction), the latter refers to the process undergone by learners who are exposed to language in a natural environment, which is then acquired unconsciously and intuitively (through social contact). Due to terminological inconsistency in the literature, SLA comprises both processes, and is therefore generally used in a broader sense. As the present study concerns language learning in a classroom environment, I will use the term second language learning (SLL) interchangeably with SLA (as well as I will not distinguish between learning and acquisition; unless there is no further indication) to refer to the discipline, as well as to the acquisition process itself. Furthermore, the term target language (henceforth L2) is used for the language to be learnt, and denotes foreign languages (i.e. languages that are not learnt in the target country, e.g.: English as a foreign language taught in German schools), as well as second languages (which are acquired within the target culture). Note that second should not be understood literally, as it can refer to any other language learnt posterior to the acquisition of the native language (or of two native languages, in case of bilingualism). SLA here includes both the study of foreign and second languages. (cf. Gass/Selinker 2001:5)

As SLA is a multi-faceted academic field, the underlying intentions of the various studies are diverse, and reveal the interdisciplinary character of the subject, as it is based on knowledge from linguistics, psychology, psycholinguistics, neuro-psychology, social and educational science, and didactics. Edmondson and House (2000:13-16) further differentiate between a number of academic fields which, to a varying degree and with different objectives, deal with language learning and are therefore closely related, i.e. applied linguistics, research in language teaching and learning (similar to applied linguistics, covering a very extensive field of theoretical and empirical methods in order to approach the issue comprehensively), foreign language didactics (practice-orientated with a strong tendency to take the teacher’s perspective), and SLA. Currently, attempts are being made to provide an institutional ground for a unified collaboration of these disciplines.

Studies in SLA tackle a number of questions, such as: ‘How does the learner process the language input which he/she is exposed to?’; ‘To what extent is learner language systematic (e.g. by questioning if there could be universal patterns of acquisition sequences in SLA, similar to those which have been found in FLA; cf. Dulay and Burt 1974; Krashen 1982) and to what extent variable?’; ‘Which learner-internal factors play a role in the process of language learning?’ and ‘To what extent is second language acquisition similar or different to the acquisition of the native language?’; etc. Within this multitude of research interests, the present study belongs to a range of papers which, from a linguistic point of view, explicitly focuses on one specific grammatical phenomenon in order to find indications of native language influence.

The following section provides a concise description of major theories in the investigation of second language acquisition.

2.1 History and theories of SLA

Influential psychological and linguistic schools of the respective times essentially determined the history of SLA. The strongest impact on theory formation in SLA was provided by the nature vs. nurture debate, which was originally concerned with first language acquisition: on the one hand, researchers (especially in the 1950’s and 60’s) account for the environment- influenced, individual-external factors such as social behaviour as the determinative element of language acquisition; on the other hand, psycholinguists argued for an innate language faculty which is said to be responsible for language acquisition (this view is largely attributed to the work of Noam Chomsky). Note that none of these approaches attempted to explain language acquisition completely one-dimensionally and isolated from other insights, but nevertheless their focus was eminently on one of the two aspects. (cf. Mitchell and Myles 2004:34f.)

2.1.1 The Contrastive Analysis Hypothesis

Since Skinner’s Verbal Behavior (1957) the prevailing view of learning as behaviour emerged. Behaviourists (such as Watson 1924) - while investigating learning processes - relied exclusively on what was observable. The Contrastive Analysis Hypothesis (henceforth CAH), which is rooted in behaviouristic learning theories, is based on the assumption that language acquisition is pure “habit formation” (Ellis 1997a:31); therefore speaking is merely seen as a habit which has to be modified according to environmental conditions. The language learner imitates and practises language in a sort of stimulus-response procedure. That means that after experiencing a certain situation (e.g. by observing how someone is asking someone else for a coffee) the learner tries to imitate the request verbally, and is rewarded by a positive reaction. If there is negative reinforcement, the learner has to rethink his utterance. (cf. Ellis 1997a:31f.)

Individuals tend to transfer the forms and meanings, and the distribution of forms and meanings of their native language and culture to the foreign language and culture - both productively when attempting to speak the language and to act in the culture, and receptively when attempting to grasp and understand the language and the culture as practiced by natives.

(Lado 1957:2)

As indicated in this quote by Lado, a central notion of CAH is that of transfer (see also section 2.2 for further details) whereby the role of the influence of the native language on L2 acquisition is significantly emphasised. In general this means (analogous to the behaviouristic view), that the learner transfers habits from L1 acquisition process (native language rules) into L2. These habits (rules), which have been acquired for L1, are seen as the major (if not the only) sources of error in second language learning as they hinder the development of new L2 habits. The central claim of early proponents of CAH (cf. Lado 1957) is based on the prediction of these errors solely by detecting where languages differ, which is also concretely expressed in its main tenet: “The greater the difference, the more errors will occur.” (Gass and Selinker 2001:72). Thus, a simple formula constitutes the basis of data analysis in the CAH tradition, that is: “degree of difference = degree of difficulty”. It was assumed that when L1 and L2 structures were similar (although “similar” is not clearly defined), positive transfer would emerge that would result in acceptable L2 utterances. On the other hand, when L1 and L2 structures differed, negative transfer (also referred to as interference) would occur. Consequently, incorrect L2 utterances were predicted, which were considered to be a product of the so-called “pro-active inhibition” (which denotes the observation that native language habits impede the learning of new L2 habits when languages differ). Thus, the central research focus of CAH is on structural differences between languages which are investigated by contrasting two languages in pairs in order to predict possible sources of learner errors. Findings from CAH research were directly used to support teaching of foreign languages by revealing possible sources of errors (which could be principally addressed in instructed SLL or in teaching material). (cf. Uzuegbu 2003:16f.)

This strong version of the CAH was unsustainable for a sufficient explanation of SLA, as empirical findings also revealed unpredicted errors which could not be traced back to transfer occurrences. Moreover, CAH was massively under attack for several reasons.

Firstly, it could no longer be denied that the sources of error were much more complex, and not solely restricted to L1 influence. Secondly, the equation of linguistic products (“difference”) and psycholinguistic concepts (“difficulty”) is scientifically untenable. Thirdly, from a linguistic perspective, an exact one-to-one comparison of languages is simply not possible, and would necessarily ignore some idiosyncratic incomparable language features, whose equivalent phenomena in other languages cannot be easily set in contrast (e.g. culture-dependent semantic value of vocabulary). Furthermore, seeing transfer errors merely as a result of language distance could also not be upheld:

A lack of contrast can also yield errors. According to this, errors occur when phenomena of the target language and their correspondents in the native language are similar, but not identical. Such errors occur more frequently as in cases of total of great differences between languages.

(cf. Edmondson and House 2000:224; translated by T.S.; emphasis in the original)

Above all, the severest failure of CAH was to completely ignore the underlying structures of the learning process per se; instead the interest was only devoted to the learners’ foreign language output on the surface level. Thus, the bulk of immense critique attacking of CAH gave rise to its weaker version (cf. Wardhaugh 1970) which was not based on the prediction of errors; rather they were analysed a posteriori. Thus, in the weak CAH, sources of error other than L1 influence were taken into account (e.g. developmental errors) which marks a turning point in the dominance of the behaviouristic-orientated psycholinguistic foundation of SLA. (cf. Uzuegbu 2003:18f.)

Nevertheless, certain aspects of CAH have never been entirely repudiated, as influences of the mother tongue always play an important role in analysing learner language.

2.1.2 Error Analysis

In the early 1970’s - as a result of its indubitable inadequacies - CAH was replaced by a new approach to second language data, namely Error Analysis (henceforth EA), which was initially introduced by Richards’s paper “A non-contrastive approach to error analysis” (1971). He

provided a typology of errors⁴ which included not only interlingual (between languages) errors, but also intralingual (language internal) as well as developmental (similar to FLA errors which reveal that the new language rules are not wholly internalised) (cf. Johnson 2001:66). This orientation towards a more scientific perspective of learner-internal cognitive processes focuses eminently on the “learner’s creative ability to construct language” (Saville- Troike 2006:37) and therefore holds the belief that learners are capable of playing an active part in SLA - in contrast to the passive role ascribed to them by behaviouristic theory. As the name suggests, EA involves the intense study of learner errors by examining their underlying sources, and therefore shifted away from mere surface observation. Language learning was now seen from a mentalist (or nativist) perspective (which represents the “innateness hypothesis”, see above 2.1) which led to a separation of didactic concerns and the study of learning processes as a study in its own right. I will succinctly comment on the related developments in psycholinguistics (see Steinberg 2002 for a comprehensive overview).

A number of SLA researchers (see, for instance, White 1996, 2003; Hawkins 2001; inter alia) transferred assumptions of the innateness hypothesis of first language acquisition to SLA: in accordance with Chomsky’s claim (in particular 1959), who was directly rejecting behaviourism (especially by addressing Skinner’s work) by arguing for a completely opposing explanation of language acquisition. Chomsky and his followers believed that a kind of mechanism called Universal Grammar (UG) was responsible for the ability of children (up to a certain age) to acquire any language to which they have access. It was suggested that UG is inherent to an innate module, the so-called Language Acquisition Device (LAD) consisting of a limited set of abstract rules which offer dichotomous choices (“principles and parameters”, see Chomsky 1981). To be more precise, a universal rule (“principle”) would be that in languages phrases always consist of basic elements (“head”) and their complements; thus a parameter could be valued by whether the complement precedes or otherwise follows the head. Applied to concrete examples, English is a head-first language, as here parts-of-speech, such as subjects, verbs, and prepositions are followed by their complements, whereas Japanese is known as a head-last language (cf. Edmondson and House 2000:138). According to this view, the LAD serves as a necessary prerequisite which has to be augmented with a sufficient amount of language input (from the respective L1 of the parents) for first language acquisition.

Nonetheless, to take up the prefatory reference to EA which is closely associated with the previously explicated mentalist framework, the work of Corder (1967, 1986) in particular has to be mentioned: his main focus was on collecting, identifying, describing, and explaining the errors of L2 learners. This approach should help to give some insight to mental processes of language learning. Thus, errors were seen as essential indicators of learner-internal developments connected with certain learner strategies of mastering language input. For instance, errors were classified according to the level of language they represent (phonological errors, syntactic errors, etc.) and further down the hierarchy with respect to really specific linguistic elements (prepositions, articles, etc.) (cf. Saville-Troike 2006:38f.). It was found that a rough division of sources could be employed; learner errors were basically based on (a) omission (leaving out necessary linguistic elements; typically articles), (b) overgeneralisation (the transfer of regular forms to irregular elements) and (c) transfer (L1 influence). Furthermore, within the evaluation procedure, more severe errors were called global errors as they turned a sentence into one which was totally unacceptable, and less problematic ones were named as local errors, as they only influenced individual sentence constituents (cf. Ellis 1997a:19f.). EA also revealed systematic developmental patterns (similar to FLA) in SLA as universal intralingual and developmental errors corroborated.

Some final remarks have to be made regarding several weaknesses of EA: Gass and Selinker (2001:86) generally criticise that EA does not comprise the whole complexity of the acquisition process by its concentration on errors which is not sufficient to explain, for example, different stages of success in SLA for learners whose language input is almost equal. Saville-Troike (2006:40) formulates the shortcomings of EA more concretely: (a) developmental and transfer errors are not easy to distinguish (e.g. Is an inflection omitted because it is not yet processed in learner language or because of it is lacking in the native language?); (b) EA owes an explanation of what actually is already acquired (the focus is only on imperfect learner utterances); and (c) sometimes learners avoid the use of certain structures (e.g. relative sentences) and make, therefore, few errors which can not be accounted for in EA. As an independent SLA theory EA was generally not fully tenable, however, it provided a sound basis for further analyses of errors.

2.1.3 The Interlanguage Hypothesis

In the field of SLA, the term interlanguage (IL) was originally coined by Selinker (1972); in the previous text the term learner language was used several times to refer to the same concept. There exist a number of other notions (transitional competence, idiosyncratic dialect, interim language, etc.) which derived from diverse theoretical contexts and are, therefore, not exactly congruent with Selinker’s term. As “a separate linguistic system” (Selinker 1972:214), IL is inherent to the second/foreign language learner, and does not represent a deficient version of the target language. It is rather a dynamic internal grammar and highly individual (nevertheless, IL is also systematic across learners). Ellis and Barkhuizen (2005:54f.) list the central premises of IL which follow in a condensed overview:

-IL comprises implicit (unconscious) linguistic knowledge and, in this sense, resembles the system of the native grammar. IL becomes apparent in learner’s L2 use.

-IL is “permeable” (IL can easily be accessed by new L2 input and internal processes of hypothesis testing.), “transitional” (IL represents only a stage of development at one point in time as it perpetually undergoes restructuring.), and “variable” (learners may employ two or even more hypothetical rules for one grammatical structure which are in some way competing which each other at any one stage of development.)

IL is seen as “the product of general learning strategies” (for example L1 transfer) which can be complemented by communication strategies (to compensate missing L2 knowledge while performing, e.g. by paraphrasing). ILs are likely to succumb to fossilisation (stagnation in development).

Following Ellis’ (1997a) view of this concept, it has to be stated that IL leaves several essential questions unanswered, for example, What, precisely, are the underlying reasons for the learner’s decision (even if unconsciously) either in favour of one of his/her hypotheses for a grammatical structure, or against it? Some special interest closely related to the IL hypothesis was concerned with the question of a universal sequence in language acquisition.

As already indicated, there are some researchers who based their investigations in SLA rather fundamentally on the Chomskyan theory (cf. Dulay and Burt 1974) and thereby almost entirely equated L1 with L2 acquisition. This concept is known as the Identity Hypothesis (or Creative Construction Hypothesis) and was implemented by a range of morpheme order studies by which researchers sought for a natural order in the grammatical development in SLA. Initial findings which identified similar developmental patterns as found in comparable studies of acquisition order in FLA (cf. esp. Brown 1973) gave rise to the claim that both the acquisition of the first as well as of the second language must be equal. However, as there are serious similarities but also undeniable differences between FLA and SLA⁵ ; and as the study of single morphemes limited to the English language was insufficient to provide serious evidence of the claim, the strong version of this approach was soon rejected (cf. Schloter 1992:10).

2.1.4 The Monitor Model

Another approach to SLA in the mentalist tradition was proposed by Stephen Krashen and was based on the following five hypotheses: (a) the acquisition-learning hypothesis which has already been elucidated above. The monitor hypothesis (b) says that the acquisition causes fluency of learner utterances, whereas what is learnt functions as a monitor of the utterances that have already been produced by the acquired system. The natural order hypothesis (c) accounts for a predictable acquisition order. Furthermore the learner needs comprehensible input⁶ as proposed by the input hypothesis (e). The affective filter hypothesis (f) asserts that the emotional state of the learner is essential for the access of this input, for instance, learners with low self-esteem, who would be seen as being not conducive to SLA, are supposed to have a “high” filter which will hinder the input reaching the brain. These proposals which form the monitor model remain widely vague, and are seen as being principally flawed. In spite of the lack of empirical evidence, Krashen’s model had a major impact on language teaching, at least in the USA. (cf. Mitchell and Myles 2004:44-49)

Apart from the theories drawing on the Chomskyan view of language which are outlined above, a number of further studies have accounted for different UG-based approaches concerning varying degrees of access to UG (also no access) for adult learners; I will not comment on them in particular. Since the linguistic theory of Universal Grammar is highly questionable itself, related studies in SLA⁷ automatically suffer the same fundamental inadequacies that shall be briefly pointed out below.

There is no substantive evidence from neuro-linguistic experiments indicating which parts of the brain are exactly responsible for language acquisition; thus, it cannot be clearly postulated that there is only one module (LAD) concerned with language processing and thereby excluding other cognitive capacities (cf. Edmondson and House 2000:148). Another controversial issue concerns the inability of UG to explain language in terms of a variety of its components and, not as it does, solely by examining single units (e.g. syntax). Furthermore, UG-based accounts are almost exclusively interested in linguistic competence, whereas facets of language use are neglected to a great extent (cf. Mitchell and Myles 2004:91-94).

The underlying point of view in this study as regards the nature of second language acquisition derives from approaches that mainly shift the learner into their focus of observation (cf., for instance, Dörnyei and Skehan 2003). So, amongst others, the so-called constructionist view appears to be of great persuasiveness, as language here is seen as not being processed independently of other cognitive processes. It is rather part of “a complex processing system that deals with linguistic information in similar ways to other kinds of information” (Mitchell and Myles 2004:97).

2.2 Reappraising transfer

The early view of transfer in SLA as proposed by CAH is not applied as the basis of the notion implied in this study. Transfer here is not regarded as the exclusive source of errors in L2 output; rather, it is part of a complex cluster of factors (e.g. learner-internal variables, social factors, quality of input, certain communicative and other learning strategies, etc.) Detached from the behaviourism-based connotation and from the mentalists’ ignorance, investigating transfer, or cross-linguistic influence (cf. Odlin 1989; Gass and Selinker 1992), is a common way to account for learner error, and therefore for a core strategy which is for some reason applied by the learner to process the L2.

As “there does not yet exist any comprehensive theory of language transfer”, and since “understanding whatever underlies the very real effects of transfer remains elusive” (Odlin 2005:478), the discussion of a revised concept of transfer cannot be really profound, but merely suggestive. Kellerman (1983 and 1986), for instance, argued for the claim of a conscious learner perception on what is transferable from the native language. Elements which are highly language-specific (i.e. the concept of “markedness”, described in detail in section 3.3), he claims, are perceived (by the learners) as being rather less likely to be transferred. Elements which are common in a language and frequently used are much more transferable, however. Applied to the context of this study, this means: German speakers are more likely to transfer a structure as Heute war Peter nicht in der Schule into English (which would result in Today was Peter not at school) than English learners of German doing the same the other way round. Admittedly, it is somewhat questionable if this theory, which was based on a study of lexical elements, can be simply applied to syntactic issues. But nevertheless, it implies a basically convincing argument.

To sum up, transfer can be seen as an essential cognitive strategy amongst a number of other learning strategies (metacognitive, cognitive, and social or affective strategies; cf. Mitchell and Myles 2004:106). What cannot be answered is: ‘When exactly and why does the learner use this strategy?’

3 Towards an understanding of inversion

In English⁸ and (more frequently) also in German, the term “inversion” applies to a considerable number of sentence structures and within its own structural bounds also covers a wide range of subcategories “as an umbrella term” (Kreyer 2006:5) of which clear- cut distinctions are not found to be standardised, at least in the relevant literature on inversion in English (cf. Green 1980; Drubig 1988; Levine 1989; Birner 1996; Dorgeloh 1997; Chen 2003; Kreyer 2006; inter alia). Apart from the fact that inversion structures are internally variable and interchangeable with respect to their concrete sentence constituents within a restricted set of possibilities they can be unified by sharing at least one major structural property - the reordering of subject⁹ and verb. Consider, for instance, the following English (1)-(4), and respectively German (5)-(8) sentences:

(1) Where did you buy your new car?
(2) Under no circumstances will she leave.
(3) “Give me the keys?” said Peter.
(4) On a bench sat little Tom.
(5) Wo hast du dein neues Auto gekauft?
(6) Unter keinen Umständen wird sie gehen.
(7) „Gib mir die Schlüssel?“, sagte Peter.
(8) Auf einer Bank saß der kleine Tom.

These are just a few examples in relation to the bulk of existing constructions which are known as inverted verb-second structures (cf. Hawkins 1986:161-194). As can be seen in

(1)-(8), each sentence structure is dominated by the inversion of subject and verb (whereas here “verb” [in English and in German examples] refers to finite forms only, e.g.: wird [verbfinite] sie [subject] gehen [verbinfinite]) with a preposed¹⁰ constituent in initial position (e.g.: unter keinen Umständen) which can (but does not necessarily has to) trigger this syntactic structure. So, a XVS pattern is established (cf. Hawkins 1986:161). Since the present work is exclusively concerned with two special subtypes of inversion in declarative sentences, namely negative and locative inversion, I preclude inversion in wh-questions, as in (1) and (5), as well as quotation inversion, as in (3) and (7), from the following explanations. I will distinguish between the remaining inversion types in section 3.1. Attempting to give an adequate and necessarily concise definition of inversion in English a brief outline of the most essential formal criteria of what inversion is, which have been selectively drawn from the literature, is provided: first, inversion constructions “display non-canonical word order, and all have grammatical counterparts […] whose truth conditions are identical” (Green 1980:582) (e.g. On the left wall hung a famous painting. vs. A famous painting hung on the left wall.). More precisely Birner says that “[a]n INVERSION is a sentence in which the logical subject appears in post-verbal position while some other, canonically post-verbal, constituent appears in clause-initial position” (1996:12, her emphasis). Furthermore, inversion as being the “reversal of the subject and the verb […] represents a deviation from the norm” and thereby a “norm-breaking” word order alternation (Dorgeloh 1997:1). The latter fact is particularly interesting when it comes to comparing inversion as a divergently marked word order phenomenon in English and in German (cf. section 3.3).

Subsequent to this introductory section on inversion, a rough (literature-based) taxonomy of inversion types is aimed for, distinguishing the two types to be investigated in this study from the remaining ones. Various studies on inversion in the English language, which approach the issue from different linguistic backgrounds, give varying accounts on what inversion actually constitutes. A further section is devoted to review the insights resulting from these studies. Seeing that in the present study predictions are made to find L1 influence on the acceptability of English inversion constructions by German learners, eventually it seems to be useful, if not necessary to provide a contrastive analysis of inversion in English and in German.

3.1 Inversion types in English

Despite the fact that most of the authors of works on inversion provide a descriptive overview of inversion types for their readers, we cannot refer to these approaches at large as a unified typology. Accounts frequently differ in fundamental issues, such as what exactly actual inversion types are, and what are just related phenomena¹¹ ; boundaries are blurred in this area.

However, an approximate uniformity is achieved with the general division between full (verb) inversion (henceforth FI) and subject auxiliary inversion (henceforth SAI)¹², which is also known under the term “subject-operator inversion” (cf. Quirk et al. 1992:1381). As suggested by their names, SAI involves the reversal of the subject and the first auxiliary of the verb phrase, whereas FI stands for all constructions in which the full verb phrase (i.e., the full (lexical) verb or copular be) is followed by the subject. Compare example (9) as an SAI instance and (10) exemplifying FI:

(9) Never before has Mary felt so lonely.

(10) Attached to the wall were several posters of his idol.

Although a highly contentious issue, SAI is a rather scarcely studied phenomenon in contrast to FI at least in empirical approaches and often excluded even from extensive studies on inversion; if it is not denied as an acceptable inversion type completely. Nevertheless, there are linguists who defend the right of SAI to be wholly accepted as a type of inversion (cf. Hawkins 1986; Dorgeloh 1997; Chen 2003; inter alia). Various subtypes are covered by SAI. Hawkins (1986:168-171)¹³ names three types: (a) inversion in wh-questions (cf. above, ex. 1), (b) negative inversion (cf. ex. 9), and (c) inversion triggered by so in initial position (e.g. So bad was his health that the doctors could no longer help him. [Hawkins 1986:169, ex. 11.24b]). Dorgeloh proposes a number of additional SAI categories (see 1997:26-29 for a complete listing). However, particularly interesting for my investigation is her statement on “negative inversion” (NegInv): in general, this SAI type is semantically triggered after front-shifted negative (e.g. never) and restrictive (e.g. hardly, rarely, only, under no circumstances, etc.) adverbs and PPs, and also constitutes an obligatory sentence process (cf. 28; but note opposing arguments below [section 3.2]). If the initial constituent which contains such a restrictive or negative element is not followed by SAI (i.e. it is instead integrated into an XSV pattern) its meaning is actually not negative and can be paraphrased in a non-negative sense. In König (1988) elements which trigger negative inversion are divided into overtly negative (e.g. not, no, never etc.) and implicitly negative expressions as barely, seldom, few, etc. The latter trigger NegInv only in sentences in which their argumentative value is clearly negative, since these expressions basically can have a positive information value. Consider example 11:

(11) a. The lesson was barely finished, in fact it wasn’t finished.

b. The lesson was barely finished, but it was finished.

Here, two possible meanings can be identified, thus the semantics of suchlike expressions is highly dependent on context. Moreover, inversion is only caused when the whole clause is negated by the appropriate adverbials; i.e. initial elements as negated conjuncts and disjuncts (e.g. nevertheless , not surprisingly) induce merely local negation (when the negation does not extent - beyond a certain phrase - to the whole clause), and therefore do not trigger SAI (cf. Erdmann 1988).

A wider range of subtypes is covered by FI. A common parameter in classifying FI types is presented by the head of the initial (preposed) phrase involved. Dorgeloh distinguishes five types (1997:23-26):

(a) AdvP-inversion: Now is the time for the first of these omissions to be rectified. (24, ex. 17, always her emphasis)

(b) PP-inversion: With greater individual freedom should come greater individual responsibility. (24, ex. 22)

(c) VP-inversion: Gone are the days when Europe’s monopoly carriers would fix prices […]. (25, ex. 23)

(d) AdjP-inversion: Conspicuous among the losers was Britain, which went in with a list of

71 items […]. (25, ex. 27)

(e) NP-inversion: An exception to this rule are the wealthy merchants. Ministers, and senior Government officials who have invested in cattle. (26, ex. 28)

All inversion types (under a different heading), except SAI types, offered by Hawkins are completely included in this classification. A particular type which he calls “Inversion after Locative Phrase Preposing” (1986:173; cf. also Green 1980) often finds special reference in various studies on inversion, and is the second inversion type (cf. ex. 12) under investigation in my study.

(12) Into the room came John’s little brother.

Not surprisingly, the terminology is also inconsistent here: I prefer the term “locative inversion” (LocInv) to terms like “stylistic inversion” or “focus inversion” (see the following section for further details) as this is the one used most frequently (e.g. Bresnan 1994). Unlike negative inversion, locative inversion is restricted to a limited set of intransitive verbs¹⁴ (see Levin 1993:92-94 for a semantic grouping of such verbs). In general, LocInv can be subcategorised to the PP-inversion type (though not exclusively: see Birner 1996:33) whereas the term “locative” can be semantically extended from denoting locations and directions to other, for instance, temporal and more abstract locative domains (cf. Bresnan 1994:75). As will become clear below, this aspect is of particular significance for the concrete learner evaluations in this study. In the following, some profound investigations on the nature of, above all, locative inversion shall be illuminated.

3.2 Theoretical preliminaries

The most current comprehensive introductory overview of syntactical, functional, and other relevant accounts on inversion in English can be found in Chen (2003:7-42). Since a detailed review of the numerous studies on inversion would go beyond the scope of this study, I will summarise his main arguments (still not withholding different voices).

As for functional accounts, including those from generative background (cf. for example Bresnan 1994), inversion has been predominantly treated in relation to “focus”¹⁵, i.e. the inverted subject NP which is hence preceded by the verb is seen as the focus of the sentence (apart from the occurrence of differences in determining the precise nature of “focus”). Opposing works (cf. Levine 1989), however, consider the preposed constituent as being focused. Amongst those, some studies presuppose, from a similar perspective, that there is in fact focus on the subject NP but that it undergoes defocusing if it is taken out of its preverbal position (cf. for instance Erdmann 1990). Chen argues that this “focus-based approach” (including all accounts on focus, also contradictory ones) is much more complex in contrast to the purely syntactic approach for it includes semantic, pragmatic, and syntactic aspects equally (2003:12).

These syntactic accounts (mainly in the generative tradition) principally discuss three major issues: firstly, if or if not inversion can occur in embedded clauses; secondly, if or if not inversion is applied because of stylistic reasons; and thirdly, if the preposed constituent can or cannot be seen as the subject of the sentence. Pitfalls of syntactic considerations are due to isolated examinations of only single aspects of inversion, and that they have been proven “either incorrect or inadequate” (2003:10); whereas functional “focus-based” accounts are reproached for being divided by lacking a uniform definition of focus.

Other functional approaches offer studies concentrating on “vividness” as the basic ground of inversion constructions, which shed more light on the nature of inversion. In addition to Chen, Dorgeloh states that “vividness“ in inversion structures can best be described by its “presentative” function (cf. Dorgeloh 1997:44), i.e. inversions “present something on the immediate stage (bring something literally or figuratively BEFORE OUR PRESENCE” (Bolinger 1977:94, emphasis in the original). According to Dorgeloh’s analysis, vividness accounts do not imply one given focus but rather a “bi-focal” marking of inversion constructions (1997:43, cf. also Drubig 1988:85). That means, a “scene” is visually laid before the addressee’s mind, and thus the inversion structure uses “lexical means” (Dorgeloh 1997:44, her emphasis) to lead the addressee’s attention to a special (existing) object within the boundaries of a local dimension in his environment on which he has to focus on. Consider example (13):

(13) Under a table sat the crying child.

As can be seen from this example, under a table constitutes - on a physical level - a spatial framework, a referential point for a “scene” where an object of the addressee’s environment, namely the crying child, can be identified. Reprehensible with this presentativefunction-based view is the restriction to locative types of inversion only¹⁶.

A supposedly more promising account, Chen claims (cf. 2003:19), would be the “information-packaging” account presented by Birner (cf. 1996:77-83; 137-139)¹⁷. She argues that inversion depends on the “linking [of, T.S.] relatively unfamiliar information which is relatively familiar (typically evoked or inferrable) in the current discourse” (77). That is, Birner applies the given/new principle which is known from “focus” analyses to the function of inversion. The notion of “givenness”¹⁸ (which denotes the part of information in a sentence which is already known to the hearer/reader from the previous conversation/text) is extended in that it is treated by differentiating between a range of types of givenness with regard to aspects of salience. A dichotomous distinction becomes rejected in favour of ranking NPs (basing on corpus data) on a scale in relation to discourse-familiarity.

[...]

---

¹ Also referred to as linguistic interference, language transfer and native language influence.

² This wording refers to a high degree of complexity of the respective element (e.g. extended by a variety of modifiers).

³ Actually, the origins of the systematic study of language learning can be traced back to the end of the 19th century (cf. Edmondson and House 2000:16).

⁴ Within EA errors are explicitly distinguished from mistakes, as the former reflect gaps in learner language (specific knowledge of L2 rules is missing), whereas the latter show “occasional lapses” in learners’ performances (Ellis 1997a:17)

⁵ In contrast to children in FLA, L2 learners already possess the necessary cognitive capability, as well as metalinguistic awareness, and knowledge of linguistic categories which facilitate language learning. Normally the native language is fully acquired and does not end in stagnation at a certain level of proficiency

(“fossilisation”) as it is typical for most of the L2 learners. Furthermore, there are no transfer errors in FLA. Nonetheless, studies showed that both acquisition processes depend on similar underlying developmental sequences. (cf. Lightbown and Spada [2006:92])

⁶ Input is comprehensible when it is just one step beyond the developmental level of the learner (in terms of syntactic complexity). If input is too simple or too complex it would not be germane for the acquisition process.

⁷ It should be noted that recent SLA studies in the UG tradition mostly refer to Chomsky’s Minimalist Program (1997) which constitutes an enhancement of the parameter and principle theory. The most striking change is that parameter setting here depends no longer on syntax, but on the lexicon. (cf. Gass and Selinker 2001:188)

⁸ All instances of this wording in the text comprise both British and American English, as there is no significant difference concerning the occurrence of inversion structures in these varieties (cf. Dorgeloh 1997:135-139).

⁹ “Subject” is understood in a traditional sense as the grammatical subject.

¹⁰ To remain terminologically consistent throughout the text the terms “topicalisation” and “topicalised” are used in a broader sense and are interchangeable with “preposing” and “preposed”. Thus, these notions do not specifically concern those particular constructions which also cause a rearrangement of sentence constituents but do not entail the inversion of subject and verb but instead form XSV patterns (e.g. leftdislocation: The blue dress, I really want to buy.).

¹¹ I therefore abandon highly controversial constructions as quotation inversion (cf. Birner 1996:20-23); there- insertion and preposing (cf. Kreyer 2006:11-16); left-dislocation and extraposition (cf. Quirk et al. 1992:1310 and 1391-1395) from the following typology. As my investigation concerns inversion in simple declarative sentences only, inversion in other sentence types is excluded as well.

¹² I am aware of views which subsume subject auxiliary inversion under the category of related phenomena of (full) inversion for their allegedly functional insufficiencies (cf. for example Birner 1996:18-20). Suffice it to say that, according to Dorgeloh (1997:21), SAI in declarative clauses incorporates (like FI as well) a word order choice and therefore “has wider discourse relevance”.

¹³ Hawkins’s classification relies to a great extent on a transformational description by Emonds (1976).

¹⁴ This does not apply to preposed constituents containing participial (transitive) verb forms (e.g. Throwing the ball was Sue’s little brother.). Furthermore and being directly relating to the inverted verb, there are also some passivised transitive verbs (seat, find, place, etc.) which allow locative inversion (cf. Bresnan 1994:78).

¹⁵ Quirk et al. refer to the notion of “focus“ as the new information (in contrast to the given information) of a message “[…] which is most neutrally and normally placed at the end of the information unit”(1992:1361). Amongst linguists studying inversion, this is but one interpretation of “focus” as it also seen in regard to

intonation, syntactic presentation, or argument structure (cf. Chen 2003:13).

¹⁶ The vividness account goes back to Bolinger (1977) and is later especially supported by Drubig (1988). An extended analysis which includes also non-locative types of inversion comes from a Cognitive Grammar framework; see Langacker (1993).

¹⁷ I will not explicitly refer to Green’s (1980) classification of six separate functions of inversion here as they can be subsumed to Birner’s account without lacking explanation.

¹⁸ Cf. Quirk et al. (1992:1360f.).