Connector Usage in Essays Written by German Learners of English: A Corpus-based Investigation Bachelor Thesis for the program “Modern Languages, Cultures and Business Studies” submitted by Cathrin Hein-Becker Department of English Language and Linguistics Justus Liebig University, Giessen Supervisor: Prof. Dr. Susanne Göpferich Cathrin Hein-Becker Auf dem Boden 4 63683 Ortenberg [email protected] Matriculation number: 3050979 Date of submission: 24.06.2013 Connector Usage in Essays Written by German Learners of English 2 Table of Contents List of figures …........................................................................................... 3 List of tables ………………………………………………………………. 3 List of abbreviations …................................................................................ 4 1 Introduction …............................................................................................. 5 2 Study design and methods of data analysis ….......................................... 10 3 4 2.1 Corpora ….......................................................................................... 10 2.2 Methodology …................................................................................. 12 Results and discussion ................................................................................ 16 3.1 Qualitative analyses …....................................................................... 17 3.2 Quantitative analyses …..................................................................... 24 3.2 Analysis of under- and overused items …......................................... 29 Pedagogical implications and outlook …................................................... 34 Works cited ….............................................................................................. 41 Erklärung zur Urheberschaft …................................................................ 44 Connector Usage in Essays Written by German Learners of English 3 List of figures Figure 1. WordSmith Concord …............................................................................. 13 Figure 2. The DEE transfer model …....................................................................... 14 Figure 3 Frequencies of all connectors used by German learners …........................16 Figure 4. Comparison of the use of the three connectors in IL and NL ………..…. 24 Figure 5. Comparison of the use of the three connectors in ILa and Ilb …............... 26 Figure 6. Comparison of the use of the three connectors in TL and NL ….............. 27 Figure 7. Comparison of the use of the three connectors in IL and TL ………....... 28 Figure 8. Overview of the use of the three connectors in the four varieties …........ 28 Figure 9. Underused items in the learners’ texts compared to TL and NL .............. 29 Figure 10. Overused items in the learners’ texts compared to TL and NL ……...... 32 Figure 11. Connectives of result in Jordan (1999: 62) …......................................... 34 Figure 12. Use of discourse markers in Burns&Smallwood (1990: 110) …............ 35 Figure 13. Comparison of quantifiers in NS and NNS writing in Rundell (2007)... 37 Figure 14. Example for a corpus-based DDL exercise ……………………............ 38 List of tables Table 1. Overview of the corpora used in the study …............................................. 10 Table 2. Error types in the connector usage of German learners of English …........ 18 Connector Usage in Essays Written by German Learners of English List of abbreviations BNC British National Corpus CA Contrastive Analysis CIA Contrastive Interlanguage Analysis DDL Data-driven learning DeReKo Deutsches Referenzkorpus EFL English as foreign language ELT English language teaching ICLE International Corpus of Learner English ICM Integrated Contrastive Model IL Interlanguage L1 First language / Mother tongue L2 Second language NNS Non-native speaker NS Native speaker SL Source language SLA Second language acquisition TL Target language 4 Connector Usage in Essays Written by German Learners of English 5 1 Introduction In our global and increasingly interconnected world, the knowledge of foreign languages, especially English, is of growing importance. Therefore, making the imparting of English language skills more effective has become a focal point of interest in applied linguistic research in the last decades. With my bachelor thesis I seek to contribute to the investigation of Second Language Acquisition (SLA) by analyzing the use of connectors in essays written by German learners of English. Connectors or conjuncts have the function of “signal[ling] logical or semantic relations between units of discourse” (Altenberg/Tapper 2008: 80) and hence significantly add to the coherence and comprehensibility of a text, aspects of writing even advanced learners and native speakers typically struggle with. Since they are used to different extents and in different ways in English and German, it can be assumed that learners will have problems employing them, especially if transfer is at play. This hypothesis is supported by the fact that transfer from German has been shown to affect English text production by German learners of English with regard to connector usage (cf. Granger&Tyson 1996:23) as well as other aspects of the language negatively (cf. Nesselhauf's analysis of the use of collocations; Nesselhauf 2004: 235). Interestingly, other studies have challenged the negative influence of transfer on connector usage in EFL writing, at least for learners from first language backgrounds other than German (cf. Altenberg/Tapper 2008: 81). There are indeed reasons to believe that the influence of transfer in EFL writing is not as significant as hitherto assumed. This thesis examines these reasons. As will be shown, German learners of English sometimes use connectors in their L2 inappropriately. The analysis of German learners' connector usage is expected to reveal problems with regard to the choice of connectors, their position in the sentence, the semantic relationships they express as well as their register appropriateness. Similar observations have already been made in previous studies (cf Altenberg/Tapper 2008, Leńko-Szymańska 2008). However, the same studies also suggest that these problems are not only due to transfer from the L1. Thus, other possible explanations for the incorrect or inappropriate connector usage will have to be found. One hypothesis is that the use of connectors as it is presented in textbooks and taught in the EFL classroom is incorrect or misleading (cf Gilquin: 24f.) If this hypothesis can be confirmed, the findings of the study are of immediate relevance to Connector Usage in Essays Written by German Learners of English 6 the design of teaching methods and materials and will ideally make suggestions for possible enhancements. In contrast to most previous studies with similar objectives, the approach taken will not only be quantitative in character, that is investigate frequencies and over- and underuse of connectors, but add a brief qualitative analysis within the realms of possibility of a paper of this scope. The material to be investigated will be taken from three corpora: The essays by German and Spanish learners of English are taken from the respective subsections of the International Corpus of Learner English (ICLE). The native English control corpus is a selection of non-fiction texts from the British National Corpus (BNC). Finally, the native German control corpus is taken from the Deutsches Referenzkorpus (DeReKo). The samples are of roughly the same size (approximately 100,000 words) and contain non-literary writing. Both the German and the Spanish learners are university students. The native texts, on the other hand, are taken from various non-fiction and newspaper articles and represent standard formal language1. The investigation of the SLA process on the basis of learner corpora has come to be known as learner corpus research. This relatively young branch of linguistics dates back to the late 1980s, when EFL researchers recognized the benefits of computer-readable collections of learner production. Learner corpus research links corpus linguistics and second language research and has the major advantage of facilitating large-scale quantitative analyses of authentic, natural language that can be used to gain insights into the SLA process and improve foreign language teaching (cf. Granger 2002: 4). Before the availability of learner corpora, language learning was investigated on the basis of the contrastive analysis of the source and the target language; or by error analysis (cf. Hasselgård/Johansson: 34). While these approaches have provided significant insights into possible causes of difficulty, they relied on intuitive or elicited data and were limited in scale and range (cf. ibid: 36f.). The main shortcoming of these early approaches to SLA research, however, was that the learner output, which is in fact the ultimate interest of foreign language teaching, had at best a peripheral position in the picture (cf. Granger 2002: 6). The approaches rather concentrated on describing the target language, characterizing the learner, and enhancing instruction (cf. ibid.). Not until the second half of the 20 th century have linguists recognized the importance of studying learner language as a 1 The problem of choosing the native norm for the reference corpora will be dealt with in detail in Section 2.1. Connector Usage in Essays Written by German Learners of English 7 separate linguistic system in its own right. Selinker (1972) coined the term interlanguage to denote this system, which is “based on the observable output which results from a learner’s attempted production of a TL norm” (Selinker 1972: 60). He identified five processes which are supposed to influence and shape this system: language transfer, transfer of training, strategies of second language learning, strategies of second language communication and overgeneralization (cf. ibid.: 61). As the learner tries to reach native-like proficiency, his or her interlanguage constantly develops under the influence of these processes. Hence, the interlanguage is of great interest to linguists interested in SLA processes. With the help of learner corpora, it is now possible to investigate interlanguage on a large scale. Computerized learner corpora are therefore a great advancement for the study of interlanguage. One pioneering project was the compilation of the International Corpus of Learner English (ICLE) coordinated by Sylviane Granger, from which the samples of German learners' essays for this study are taken. It was Granger's aim to move away from mere error analysis and intuitive hypotheses on the basis of contrastive analyses of two languages to a more integrated investigation of the learning process by examining the learner's actual production. She developed the Contrastive Interlanguage Analysis (CIA), a methodological approach to carry out “quantitative and qualitative comparisons between native (NS) and non-native (NNS) data or between different varieties of non-native data” (Granger 2002: 12). This twostage comparison makes it possible to identify non-native features of learner writing and make speculations as to whether possible problems are L1-dependent or developmental (cf. ibid.: 13). For this purpose, Granger (1996) developed the Integrated Contrastive Model (ICM). It consists of two components: The contrastive analysis (CA) component, which compares the native language and the target language; and the contrastive interlanguage analysis (CIA) component, which compares the native language and a non-native variety as well as different non-native varieties of the same language (cf. Granger 1996: 46f.). The ICM is especially suitable for the investigation of L1 transfer in the language learning process. The CA component allows for a predictive hypothesis as to whether transfer from the mother tongue is likely to occur by determining the degree of similarity between the two languages. This hypothesis alone would not suffice to make legitimate assumptions as to the likelihood of negative transfer, since discrepancies between L1 and L2 do not necessarily lead to errors (cf. Gilquin 2002: 7). Therefore, the CA component is Connector Usage in Essays Written by German Learners of English 8 complemented by the CIA component, which investigates the interlanguage and returns to CA to explain L1-specific errors. If the comparison of mother tongue and target language cannot account for these errors, other factors such as developmental errors or teaching-induced factors might be at play. Numerous researchers adapted Granger's approach in their own research and carried out a multitude of studies within the framework of learner corpus research. Most of them focused on the use of specific lexical or syntactical elements, as the present study does. These studies were only quantitative in character and led to ambiguous results regarding the role of transfer (cf. Leńko-Szymańska 2008: 95). Altenberg & Tapper (1998), Granger & Tyson (1996), and Leńko-Szymańska (2008) all observed instances of over- or underuse of certain connectors; and they all attribute these at least partly to transfer from the mother tongue. Similarly, Nesselhauf (2003), who investigated the use of collocations by German learners of English, found that 56% of the mistakes made in the use of collocations can be traced back to L1 influence (Nesselhauf: 235). There is extensive evidence that “transfer is one of the major factors shaping the learner's interlanguage competence and performance” (Kohn 1986: 2). However, the above-mentioned authors concede that transfer cannot solely be held responsible for the erroneous production of learners of English. They acknowledge the influence of other factors, above all teaching-induced effects. Granger&Tyson, for instance, find fault with misleading “lists of 'interchangeable' connectors” (Granger&Tyson: 23); and Gilquin draws attention to the “unsuitability of some pedagogical materials and/or reference tools” (Gilquin: 24f) . In spite of legitimate counter evidence, the negative influence of L1 transfer on the written production of learners of English remains the prevalent doctrine. Researchers working with learner corpora have not reached consensus on other possible explanations for the problems they observed; and the focus in learner corpus research seems to have shifted towards investigations of the pragmatic and discourse level with the emergence of the research-field of contrastive rhetoric (cf. Herriman & Boström Aronsson 2009: 118). By analyzing the use of connectors by German learners of English both quantitatively and qualitatively, I am going to revisit the problem of transfer in my bachelor thesis and show that, although it might take place in the written production even of advanced learners of English, it is not the only reason for the misuse of connectors. Connector Usage in Essays Written by German Learners of English 9 In order to do so, I will conduct a corpus-based study of the three connectors used most frequently in a corpus of essays written by advanced German learners of English. The three corpora used for these investigations will be presented in Section 2.1. I will analyze qualitative errors the German learners make when using the connectors in questions and compare the frequency of the connectors in the texts written by German learners to that by Spanish learners and native speakers of English, as well as to the frequency of their correspondents in native German published writing. My analysis will be guided by the steps of Gilquin's (2002) Detection – Explanation - Evaluation (DEE) transfer model, which will be explained in Section 2.2. The results of the analyses will be presented and discussed in Section 3. Section 4 addresses some pedagogical implications of my findings and of learner corpus research on the whole and show how they could help enhance teaching materials and methodologies in the future. At the same time, I am going to show that there is still great potential for SLA research to be exploited with the help of corpus linguistic methodology, and I am going to name but a few possible areas of investigation that might be worth looking into in future studies. Connector Usage in Essays Written by German Learners of English 10 2 Study design and methods of data analysis The present study is based on a series of studies dealing with similar subject matter in the late 1990s and early 2000s (see Section 1). These studies investigated the use of connectors and other lexico-grammatical phenomena in a quantitative way. In response to criticism leveled against at purely quantitative studies (cf. Granger 1998: 16), I will undertake a qualitative analysis as called for by Leńko-Szymańska (2008), who expressed the need for such an analysis “in order to investigate whether linking expressions are employed to indicate real semantic relationships between ideas or whether they are used inappropriately” (Leńko-Szymańska 2008: 105). I will take a closer look at certain connectors with regard to their frequency, but also qualitative problems German learners seem to have when using them in their writing in English. Of the studies mentioned, only Leńko-Szymańska (2008) analyzed a smaller selection of connectors in more detail (cf. Leńko-Szymańska 2008: 99). She looked at the ten most frequent connectors in a total of nine samples with sizes of between 50,000 and 75,000 words respectively. This was still too broad for a detailed qualitative analysis, so the study also only yielded quantitative results. Therefore, in the present study the use of only three connectors and their correspondents in German will be investigated. This method will lead to more profound results, on the basis of which some pedagogical implications and proposals for further research can be formulated. 2.1 Corpora The data samples to be investigated will be taken from the three corpora named in the introduction. Table 1 provides an overview of these samples and their sizes. Corpus Subjects Text type ICLE-DE ICLE-ES DeReKo German learners of English Spanish learners of English German professional writers BNC English professional writers Argumentative essays Argumentative essays Published newspaper and non-fiction articles Published newspaper and non-fiction articles Size (number of words) 108,540 177,350 105,240 105,390 Table 1. Overview of the corpora used in the study The ICLE is a large compilation of advanced EFL learner writing by learners from 16 native language backgrounds. It is the outcome of a large-scale project launched Connector Usage in Essays Written by German Learners of English 11 by Sylviane Granger at the Université Catholique de Louvain in 1990. Granger's aim was to gather empirical data for the investigation of interlanguage. The corpus design was strictly controlled with regard to the type of learners (EFL, not ESL), the stage of the learners and the text type in order to make the samples comparable (cf Granger & Tyson 1996: 18). All the learners are young adults in their twenties who study English in a non-native context, and are university students of English Language and Literature in their third or fourth year, which makes them 'advanced' learners (cf. Granger 2002: 9f.). Their essays cover a variety of non-technical topics and are comparable insofar as they represent argumentative as opposed to narrative writing (cf. ibid.: 10). Granger argues that “[t]hese essays (…) are unabridged and so lend themselves to analyses of cohesion and coherence” (cf. ibid.), which makes them especially suitable for the present study. Beyond these basic design criteria, the ICLE samples feature a number of variable attributes such as sex, mother tongue background, experience in English-speaking countries and task settings. For the present purpose, the German and Spanish samples where filtered in terms of mother tongue (only German or Spanish as L1, respectively), and experience abroad (less than 6 months in order to guarantee unaltered 'non-nativeness'). The German subcorpus comprises 108,540 words, the Spanish one 177,346 words. The British National Corpus (BNC) is a 100 million word collection of samples of written and spoken British English2. It includes, amongst others, extracts from newspapers, specialist periodicals and journals, academic books and popular fiction, as well as school and university essays (cf. Burnard 2009). A sample of nonfictional published writing with a total of 105,390 words will be used in this study. Criticism has been voiced against choosing expert writing as a reference: Granger (1998: 13) argues that only corpora of the same genre and level of expertise are directly comparable. Leńko-Szymańska (2008) confirms this by arguing that “the observed differences will only reflect the disparities in linguistic systems and will not be a result of discrepancies in the level of writing skill” (Leńko-Szymańska 2008: 96). Hasselgård and Johansson (2011), on the other hand, claim that [f]rom an English Language Teaching (ELT) perspective, however, a student corpus […] may be considered unsuitable as a reference corpus because it does not represent the desired target norm for proficiency or the type of language one would like to teach […]. Thus, if the aim is to identify areas of argumentative or academic writing which learners need 2 British English was chosen as the native norm here because it is the variety of English usually used as linguistic model in EFL teaching in Germany (cf. Lenko-Szymanska 2008: 98). Connector Usage in Essays Written by German Learners of English 12 to improve, an NS corpora consisting, for example, of press editorials or academic articles may be preferable. (Hasselgård/Johansson 2011: 39) What EFL learners are usually exposed to and hence aspire to approach is indeed published expert writing (cf. Gilquin 2008: 24). Furthermore, Leńko-Szymańska (2008) points out that students of English as a foreign language usually receive more explicit writing instruction than native speakers and are thus possibly more skilled in writing techniques. This would of course distort the validity of native/non-native comparisons and pedagogical recommendations based on them (cf. Hasselgård/Johansson 2011: 55). Hence, a control corpus of expert writing seems to be a reasonable choice for the purpose of this study. The native German texts are taken from the Deutsches Referenzkorpus (DeReKo), a collection of texts compiled by the Institut der Deutschen Sprache to represent present-day German. The corpus comprises texts from a multitude of genres, including newspaper and journal articles, works of fiction, as well as scientific, legal and political texts (cf. Institut der Deutschen Sprache 2002). The samples used for this study are a selection of non-fiction and newspaper articles with a total of 105,240 words. Here, too, published expert writing was chosen as a reference for the reasons outlined above. All three corpora are accessible in electronic form on CD-ROM (ICLE) or online (BNC and DeReKo) respectively and are thus computer-readable with the help of corpus-linguistic concordance software, which will be described in the next section. 2.2 Methodology In order to investigate the use of connectors, a working definition of 'connector' is necessary. For the purpose of this study, I follow Quirk et. al.'s (1985: 631) understanding of connectors (what Quirk et al. call 'conjuncts') as 'adverbials with specific semantic roles conjoining independent units'. Accordingly, I will use their extensive list of connectors (cf. ibid.: 634ff) as a starting point for my investigation. This selection has already been proved useful in earlier studies on the subject (cf. Granger&Tyson 1996: 19f; Altenberg&Tapper 2008: 81) and allows for comparability between the results of these studies and my own findings. Connector Usage in Essays Written by German Learners of English 13 The three connectors from Quirk et al.’s list used most frequently in the texts by German learners of English will be identified with the help of the corpus-linguistic concordance software WordSmith Concord. Concord seeks a specified search item in a chosen text sample and generates a concordance display, together with information about the context in which the search item appears. Figure 1 depicts the concordance lines for the inquiry of however in the German learners' texts. Figure 1. WordSmith Concord To determine the three connectors used most frequently in the sample of German learner writing, all items on Quirk et al.’s list will be used as search items in Concord and their numbers of occurrences will be compared. The three most frequent will then be subjected to a qualitative analysis by assessing the correctness and appropriateness of their semantic function, their position and their register suitability. Subsequently, the three connectors will be used for further concordance queries in the other text samples using the same method. Once the frequencies have been determined using the corpus-linguistic software in all four samples, they will be compared in order to determine in how far the occurrences in learner writing can be explained by L1 transfer. Whereas in previous studies this was done solely by comparing the frequencies of certain connectors in learner writing with either their frequency in the mother tongue (Evensen&Rygh 1988) or in native English (Granger&Tyson 1996), my quantitative approach will compare mother tongue, interlanguage and target language. This is achieved with the help of the Detection – Explanation - Evaluation (DEE)– Model developed by Gilquin (2008). This model combines the methods of two earlier models for the investigation of transfer in EFL writing, Granger's (1996) Integrated Contrastive Model and Jarvis's (2000) unified framework for transfer research and Connector Usage in Essays Written by German Learners of English 14 comprises a total of four3 comparisons. The DEE model comprises all three language varieties involved in the SLA process (L1, interlanguage and L2) and aims to determine when transfer is at work, what its origin is and whether it is pedagogically relevant (cf. Gilquin 2008: 4). It hence “proposes a methodology which seeks to apprehend transfer in a way that takes several of its factors into account, thus striving for a comprehensive view of the phenomenon” (Gilquin 2008: 13). Figure 2. The DEE transfer model based on Gilquin (2008) Figure 2 depicts the three stages of analysis of the DEE model. In a first step (1), the learner's interlanguage is compared to his or her mother tongue (NL/IL). This is a classic case of contrastive analysis (CA). If a similarity between both can be observed, it is plausible that transfer occurs (cf. ibid.).Yet, a congruity between the mother tongue and the learner's interlanguage alone cannot account for the presence of transfer, because the learner's behavior may not be restricted to learners with his or her mother tongue but may well be common to learners with other native languages. This is why, in addition, the learner's interlanguage will be compared to the interlanguage of learners of English with a different mother tongue (ILa/ILb), that is, 3 The original model by Gilquin (2008) comprises six comparisons. In addition to the ones used in this study, she also includes a comparison of source language and translated language (SL/TrL) with the help of a parallel translation corpus. Furthermore, she adds an explicit comparison of the interlanguages by learners with the same mother tongue (ILa/ILa, cf. Gilquin 2008: 5). For convenience, the German and Spanish learners' interlanguages are each considered homogeneous varieties in this study. Where necessary, however, an intra-group comparison will indirectly be integrated in the analysis in cases where individual learners' productions stand out in the samples. Connector Usage in Essays Written by German Learners of English 15 a contrastive interlanguage analysis (CIA) approach will be applied (2). Only if that second comparison fails to reveal similarities between the interlanguages can the researcher assume to have detected L1 transfer with some degree of confidence (cf. ibid.: 14). The detection stage is followed by an attempt to explain the occurrence of transfer. This is done by assessing the “language distance”, i.e. the degree of similarity between the mother tongue and the target language (cf. ibid.: 16). Again, a contrastive analysis is conducted (3). The items under investigation in the learner's native or source language (SL) are compared to the use of their equivalents in a comparable control corpus of the target language (TL) with regard to their frequency, function and stylistic appropriateness (SL/TL). If the two languages can be proved to be similar with respect to the use of the respective connectors, the learner will most likely be tempted to transfer from one to the other (cf. ibid.). In the final stage of the DEE model, the interlanguage is compared to the target native language (TL/IL) in order to determine whether the detected instances of transfer are cases of negative or of positive transfer (4). This, in turn, is a CIA approach again. If the transferred item or structure is absent in the target language, the mother tongue influenced the learner's interlanguage negatively. Correspondingly, the production of a correct target form in the interlanguage as a result of L1 influence indicates positive transfer (cf. ibid.: 18). Having evaluated the learner's performance this way, Gilquin claims that “[g]iven the time constrains that exist in the FLT classroom, it is best to focus on negative transfer […], showing learners how to avoid it […], and let positive transfer […] make its way into the interlanguage.” (cf. ibid.). Pedagogical implications will be addressed in Section 4. In the next section, the use of three connectors used most frequently by German learners of English will be analyzed in a qualitative way. Subsequently, the three stages of the DEE model will be used to guide quantitative analyses of the connector usage of German learners of English. Connector Usage in Essays Written by German Learners of English 16 3 Results and discussion The analysis for the frequencies with which the connectors used as search items occurred yielded the results depicted in Figure 3. The figure depicts the frequencies of all the connectors that were used at least 0.5 times per 10,000 words in the German learners' texts. Since the samples differed in size, normalized values of tokens per 10,000 words will be used in what follows in order to ensure comparability. however 8,75 therefore 5,16 nevertheless 2,49 though 2,49 of course 2,49 on the other hand 1,75 for example 1,57 furthermore 1,57 besides 1,11 moreover 1,11 that is to say 0,92 by the way 0,83 on the contrary 0,83 for instance 0,83 otherwise 0,74 what is more 0,74 admittedly 0,64 at the same time 0,64 consequently 0,64 on the one hand 0,55 after all 0,55 hence 0,55 in addition 0,55 0,00 2,00 occurence per 10,000 words 4,00 6,00 8,00 10,00 Figure 3. Frequencies of all connectors used in the English essays written by German learners As shown in Figure 3, the connector which occurred most frequently was the contrastive however. The second most frequent was the summative therefore, and the third place was occupied by the contrastive nevertheless. It has to be mentioned that the ascertainment of the third most frequent connector was not as straightforward as the first two, because there were two more connectors with the same frequency. Connector Usage in Essays Written by German Learners of English 17 However, for the others, namely of course and though, it was difficult to determine in which cases they served as a connector in the sense defined by Quirk et al. (1985). Of course, for example, can be used as a resultive connector in the sense of therefore or so as in “I even was prepared to put up with certain disadvantages: I knew that the walls between the different flats were very thin and that my neighbours - an elderly couple - went to bed early: of course I would turn down the music after 8 o'clock” (<ICLE-GE-AUG-0042.2>, my italics) or as a contrastive concessive in the sense of although as in “ Of course, we know that this idea might cause difficulties in a big city, but why shouldn't they have a try?” (<ICLE-GE-AUG-0007.1>, my italics)(cf. Quirk et al. 1985: 638f.). Much more frequently (16 out of 27 occurrences) it was used as an intensifier in the English essays written by German learners, as in “In the groove of a record the sound-waves have their exact replica, the term analogue recording implies this CD players try to counter this problem with oversampling, but that is of course not as exact as analogue sound reproduction” (<ICLE-GE-SAL0009.2>, my italics). Though is called a “marginal case” by Quirk et al. (Quirk et al. 1985: 642). It only counts as a connector if it occurs in sentence-final position, as in “Having no money left at all there was nothing left to do than hitchhike to the hamlet where she said she was living. The wedding was brilliant, though” (<ICLE-GE-AUG-0082.3>, my italics). In sentence-initial or middle position, on the other hand, it is not a connector but as subordinator equivalent to although, as in “Though I was still a small child, I felt that the skill of cycling would open the world, which was at that time my village, to me” (<ICLE-GE-AUG-0056.3>, my italics)(cf. Quirk et al. 1985: 642). The German learners used though in the majority of cases (22 of 27 occurrences) in this latter sense in their English essays; that is, they did not often use it as a connector. Therefore, nevertheless was the third connector that could unambiguously be classified as such in the most occurrences. However, though will be returned to later when taking a closer look at underused items. 3.1 Qualitative analyses As regards the qualitative assessment of the use of however, therefore and nevertheless, some problems with regards to the necessity, semantic function, explicitness/implicitness of the statement and positioning of the connectors could be observed. Table 2 provides an overview of the error types, their raw frequencies Connector Usage in Essays Written by German Learners of English 18 within the total occurrence of however, therefore and nevertheless (178 occurences) and their percentages. For each error type, random representative examples from the essays written by German learners of English will be analyzed. Error type Occurrence Percentage Redundancy 6 3.37% Semantic function 5 2.81% Implicitness 4 2.25% Sentence structure 1 0.56% Punctuation 9 5.06% Table 2. Error types in the connector usage of German learners of English The fact that however, therefore and nevertheless were the most frequent connectors in the learner texts is not surprising, since their semantic function to present “contrastive matter in relation to what has preceded” (however and nevertheless) or “to introduce and item that embraces the preceding ones” (therefore)(Quirk et al. 1985: 638) is familiar to the learners because similar structures exist in their mother tongue (see the discussion of the German correspondents below). The problem seems to be that learners are so comfortable with the use of these connectors that they tend to overuse them, even when they are not necessary. One of the subjects, for instance, used however four times in a thirty-sentence essay (<ICLE-GE-AUG-0089.3>). The following examples are cases of redundancy. [1] Obviously everyone is familiar with the horrible pictures of crying mothers who lost their men and children in the second world-war. But recently there have been, and there are still fierce battles, like in Iraq or in the former Yugoslavia. However, no war is the same, as they have all different causes and objectives. So, one should ask himself the question, whether a war is justified or whether peace is a good thing at any price. There is no doubt, a war with all its consequences, has to be in any cases only the final and last solution of problems. But nobody can deny that a war is sometimes the most effective solution, when other methods like economic sanctions are useless. (<ICLE-GE-AUG0038.1>, my italics) The connector however is not exactly redundant here, but not necessary either. The statement that “no war is the same” is not in contrast to what has been said before and could therefore stand on its own. The author should have left however out in this case because in this relatively short passage, he starts nearly every sentence with a connector, which makes his style appear forced and unprofessional. Furthermore, in addition to the sentence introduced by however, he used but twice. But signals a Connector Usage in Essays Written by German Learners of English 19 contrast as well, which means that in an essay of six sentences, the author included three contrasts and hence shifted direction in his argumentation three times. Instead of creating coherence, the exaggerated use of connectors confuses the reader and reduces the quality of the text. This is also true for example [2]: [2] Even students or workers sometimes have to drive by high speed, as they are late for work for several reasons, e.g. not having heard the alarm-clock and therefore having got up too late. <ICLE-GE-AUG0027.1> It is evident that the person in the example got up too late because they had not heard their alarm. Emphasizing this causal relationship with the help of the connector therefore is not wrong, but it is not necessary either. The sentence is already complicated because of the intricate use of the infinite verb forms with the past participle, and the redundant connector worsens this complexity. In order to avoid a clumsy style due to the exaggerated use of connectors, the author should have left it out in this case. The following excerpt is another example for redundant connector usage: [3] And even though at Apuleius' time a novel seemed to be an unusual means of moral education - this was his major concern. Like Rabelais he obviously found that popular tales gave him a wider field for his descriptions of contemporary morals and manners punctuated by philosophical asides, than any more respectable literary form at his time. Therefore the entertaining frame is just an instrument for other purposes such as social satire and literary parody. <ICLE-GE-SAL0010.2> The author of this passage provides arguments for her claim that “the entertaining frame” of Apuleius' novels was “just an instrument”. The quoted passage leads to ther central hypothesis. The summative function of the last sentence is achieved through discursive strategies, and the connector therefore is not needed to clarify it. In fact, it is considered professional style to achieve this kind of cohesion without the use of connectors. (cf. Leńko-Szymańska 2008: 104) Hence, the learner actually would have produced a professional text, had they not spoiled it with the unnecessary connector. Once more, they unnecessarily clung to their 'lexical teddy bear'. This is confirmed by the fact that this learner used therefore five times in their two-page essay. There were also cases where nevertheless was used redundantly, as in example [4]: [4] Despite her social criticism the authoress nevertheless does not mean to shock her readers - on the contrary. <ICLE-GE-SAL-0010.2> Connector Usage in Essays Written by German Learners of English 20 In this example, nevertheless is redundant because the contrastive relationship between the mentioned authoress' actions and her intention is already sufficiently made clear by despite. Nevertheless and despite have the same semantic function. Hence, only one of the two is needed. The cooccurrence of conjuncts is, according to Quirk et al. (1985: 642), “stylistically questionable” and “more characteristic of loose informal talk”. Besides this redundant use of nevertheless, the learner used this connector two more times in their essay, which may be taken as a sign of overuse. The next example is a case of semantically incorrect use. The semantic relation however expresses is not appropriate in the context of this passage: [5] As we already know, Neofascism is a dark part of our history that is accompanied by cruelty, discrimination and crime. Many crimes commited [sic] in recent years go back on neofascist tendencies; tendencies that seem to be legal and becoming stronger. However, if the government does not limit the influence of neofascism we will find no way out of this problems. (<ICLE-GE-DRE-0026.1>, my italics) The sentence “if the government does not limit the influence of neofascism we will find no way out of this problems” is not a contrast to but a conclusion from what has been stated before. Therefore the connector however must be replaced by a concluding connector. Similarly, in example [6] the author used a resultive connector where he should have used an adverb of manner. [6] They'll be around you when you're not in a good mood, they're glad when you go outside for a walk with them, they say hello with their tale, the eyes and their voice "wau, wau", and therefore show you how excited they are to see you. <ICLE-GE-AUG-0059.1> What the author wants to say is not that a dog shows its excitement because it wags its tale and barks, but they show it by doing so. Hence, therefore is wrong in this context. The sentence should read “... and thereby show you how excited they are”. The learner presumably confused therefore and thereby due to their formal similarity. Another frequent error type in the connector usage of German learners of English was implicitness. This describes cases when authors use a connector that signals an explicit semantic relation such as contrast or consequence to refer to a statement they only implicitly made. Example [7] is a case in point: [7] [C]hildhood, associated with myths of unspoilt nature, frankness and freedom. Nevertheless these myths need to be corrected. <ICLEGE-AUG-0004.3> Nevertheless is a concessive contrastive like however (cf. Quirk et. al. 1985: 636). Connector Usage in Essays Written by German Learners of English 21 But while however expresses a simple contrast and is semantically coinciding with but, nevertheless is used in statements that are true in spite of something that has been said before. Thus, it does not only contrast two units, but requires an explicit counter-statement. The phrase “childhood, associated with myths of unspoilt nature, frankness and freedom” indeed implies that the myths associated with childhood have been commonly accepted for a long time, and the author wants to challenge these myths now. Hence, the semantic relationship for nevertheless is given, but it is not explicitly stated. The first unit is only a noun phrase and lacks a predicate for nevertheless to refer to. This is why nevertheless appears odd in that sentence. The author of the following excerpt also misused the explicit contrastive function of nevertheless: [8] Being an "ecologist" costs much more, more money, more nervs [sic], more time. But nevertheless they are better people! Whereelse [sic] should we start than with us? (<ICLE-GE-AUG-0033.1>, my italics). This is another case of objectionable implicitness. The author's argumentation is that pending money, nerves and time on environmentally friendly behavior, though, does not contrast with being a better person in the learner's argumentation. Rather, it is a proof of a person's benevolence. The author does not explicitly state this claim with a statement like environmentally friendly behavior takes a lot of effort but it is worth it or but it is necessary. Hence, he cannot draw an explicit contrast with nevertheless between two units the contrast between which he has only implicitly stated. A similar error is at hand in the following example: [9] Certainly this first ride had been followed by many others, and after a time I began to travel longer distances as well. At that time, I however also had to cope with several problems which cycling naturally brings about. (<ICLE-GE-AUG-0054.3>, my italics) In this example, the use of however is not correct either. The author wants to express that riding a bike is not only pleasure, but also carries problems. However, they do not explicitly verbalize that pleasure and thus cannot explicitly contrast it with the problems introduced by however. The extensive and often problematic use of however, therefore and nevertheless in the learner texts can be explained by Hasselgren's concept of 'lexical teddy bears' (Hasselgren 1994: 237). She found that learners overuse words that are familiar and that they feel safe with and cling to them even in cases where they are unnecessary or where other lexical choices would be semantically more precise, as Connector Usage in Essays Written by German Learners of English 22 well as in cases where they are actually wrong (cf. ibid). The three connectors in question can be said to be such 'lexical teddy bears' for many German learners of English because they use them extensively in their essays even when they are redundant or incorrect. They seem to think that the more connectors they use, the better the quality of their texts gets (cf. Leńko-Szymańska 2008: 105). In reality, though, the opposite is the case. Herriman & Boström Aronsson's (2009) conclusion from their comparable analysis of Swedish learner's writing in English also applies to the present study: [T]he NNS [non-native speakers][…] appear to lack sufficient knowledge of textual organization in English. This is suggested by their overuse of explicit expressions of cohesion, such as conjunctive[s] which express relationships between ideas and parts of the text, which, in turn, may result in a clumsier style in their essays. This overuse may be due to an insecurity among the NNS as regards the creation of cohesive ties when writing in a foreign language. (Herriman&Boström Aronsson 2009: 117) The exaggerated use of connectors will be returned to in Section 4. Other aspects that need to be considered when assessing the quality of connector usage are sentence structure, position and punctuation. All three are formal rather than contentrelated aspects of connectors. Since this study is more interested in the discursive function of connectors, they will be addressed only briefly. The analysis of the structure of sentences with connectors in the English essays by German learners is especially interesting because German, in contrast to English, allows for a variable word-order. The use of a connector often leads to a rearrangement of the sentence (e.g. Er ist krank, er kommt heute nicht zur Arbeit versus Er ist krank, daher kommt er heute nicht zu Arbeit). Such a rearrangement is not legitimate in English. Some learners seem to be unaware of this difference, as the following example shows: [10] It was proved that TV commercials have a large impact on our subconciousness [sic] and that we can easily be brought to by [sic] things by such psychological tricks. (…) Therefore should TV commercials be banned. (<ICLE-GE-AUG-0083.1>, my italics) Here, the learner clearly transferred the German sentence structure 'Daher sollte Fernsehwerbung verboten werden', while a restructuring of the sentence to something like 'This is why TV commercial should be banned' would have been a better choice. However, the example just quoted was the only case of such syntactical errors. English structure in sentences with connectors does not seem to be a major problem for German learners. The position of the connector was even more Connector Usage in Essays Written by German Learners of English 23 unproblematic. According to Quirk et al., the clause-initial position is the most common position for connectors. In addition, some can also occur in the middle or at the end of a sentence. However, therefore, and nevertheless can all occur in clauseinitial and middle position (cf. Quirk et. Al 1985: 643). In their cases the positioning does not cause problems because the German equivalents are used in the same positions. The learners did not once use them in the sentence-final position. Only the use of commas really posed a problem to the learners. Connectors in the sentenceinitial position are usually followed by a comma (cf. ibid.), and those in the middle position can be separated by commas if the semantic role of the connector, e.g. a contrast, should be emphasized. The learners, however, often used the connectors without commas, as in the following examples: [11] Consulting a dictionary about the different meanings of the word "queer" however, made me change my mind: maybe being queer really is a prerequisite for challenging society.(<ICLE-GE-AUG-0065.3> [12] I have all duly respect for womanly beauty, intelligence and charm and I have been known to courtsey [sic] to any female showing respect for male dominance. However I can't put up with the condescending arrogance ever more of today's liberal women apply to treat today's newly oppressed men. [13] So I can't see the argument for driving at high speed. Therefore I plead the case for introducing a speed limit on German motorways. (<ICLE-GE-AUG-0028.1>) [14] My italian [sic] cousin is not familiar with bavarian [sic] civilisation, and so he asked me why the local population stood on the tables instead of sitting on their chairs. Nevertheless he has enjoyed this exotic behaviour (...). (<ICLE-GE-AUG-0059.3>) The omission of commas is legitimate in the cases of therefore and nevertheless, where the comma is optional and would only be necessary “when the conjunct would have a separate intonation nucleus in speech or when it can be misinterpreted” (Quirk et al. 1985: 643). However, on the other hand, has to be separated by commas because the contrast it signals requires accentuation and a pause in speech. Without the comma, however could be read as an adjective, as in the sentence “As a matter of fact, however marvellous [sic] computers are said to be they simply cannot replace humans talking to one another.” (<ICLE-GE-SAL-0003.2>). The punctuation, or rather its absence, in examples [11] and [12] is therefore erroneous. Overall, qualitative problems, both formal and functional, occurred in more than 14 per cent of the instances of the usage of however, therefore and nevertheless Connector Usage in Essays Written by German Learners of English 24 by German learners of English. In addition, they often used their favored connectors where others would have been more appropriate. Hence or thus, for instance, would have been more formal alternatives to therefore. The underuse of more sophisticated alternatives will be addressed later in this section. In summary it can be said that this brief qualitative assessment of problems with the use of the three most frequent connectors in German learner writing revealed some problems with the explicitness of the connectors, as well as considerable uncertainty as to their necessity, meaning and register-appropriateness. 3.2 Quantitative analyses In a subsequent step, the frequencies of the three determined connectors were compared to the frequencies of the respective functional equivalents in German. As correspondents, jedoch, aber and allerdings for however, daher, folglich, demzufolge, somit, mithin and deshalb for therefore, and trotzdem, nichtsdestoweniger, trotz alledem, (dessen) ungeachtet and dennoch for nevertheless were chosen. Two works of reference were consulted for the determination of the equivalents: The Oxford German Dictionary and the seventh edition of the Duden. It is hardly ever possible to determine the exact equivalent of a given word in a foreign language, and translation studies have moved away from such attempts towards a more functionally oriented paradigm. Thus, some of the chosen German equivalents do not belong to the same grammatical category as the English connectors, but they serve the same function and should therefore be considered in the analysis. The chosen German words were queried in the German corpus and compared to the frequency to their correspondents in the German learner texts (Step 1 of the DEE model) with the results shown in Figure 4. 11,4 12 10 8,75 8 6 4 5,16 3,61 2,49 3,32 2 0 however vs. jedoch/allerdings therefore vs. deshalb, daher, also, folglich, demzufolge, somit, mithin nevertheless vs. Trotzdem/dennoch Figure 4. Comparison of the use of the three connectors in IL and NL IL NL Connector Usage in Essays Written by German Learners of English 25 Added together, jedoch and allerdings occurred much less frequently in native German. The difference to the frequency of the use of however in the learner texts may be explained by the problem of finding an equivalent. Where English uses however, German often does not use an adverbial connector. Instead, aber (but) is used. When looking at functional equivalence, aber has to be considered, as well. It occurred 38.67 times per 10,000 words. However, if aber is considered in the native German texts, the frequency of but in the English essays written by German learners is also interesting. It occurred 62.37 times per 10,000 words. Hence, both the comparison of the frequencies of contrastive connectors only as well as that of all items serving a contrastive function show a much higher frequency in the interlanguage than in native German. The functional correspondents to therefore occurred at a sum of 11.40 times per 10,000 words, which is not in accordance with the frequency in the learner texts, either. Even if the functional equivalents to therefore in English, namely hence, thus and consequently, are taken into consideration, connectors signaling results or consequences in learner English still only amount to 8.57 occurrences per 10,000 words as opposed to more than 11 occurrences per 10,000 words in native German. Items with this function, then, seem to be more frequent in German. The German equivalents of nevertheless, on the other hand, exhibited more similar frequencies to that of their English counterpart in the learner texts. The function of nevertheless is similar to however. The similarity of its frequency in the English essays written by German learners and in native German can thus not be explained by functional reasons, because for however the figures were very different. Nevertheless, in two of three cases, the German learners seemed to be using connectors for the same semantic relations at disparate frequencies in their interlanguage as they are used to in their native language. This suggests the unlikelihood of transfer. The IL/NL comparison is complemented in the DEE model by a comparison of German learners' IL to the IL of learners with a different mother tongue (Step 2 of the model), in order to ascertain whether possible problems are L1-specific and hence a result of transfer or if they are common to all learners. Therefore, the three most common connectors in the essays by German learners were inquired in the sample of texts produced by Spanish students. The frequencies of the connectors in these texts in comparison to the German sample are shown in Figure 5. Connector Usage in Essays Written by German Learners of English 26 10 9 8 8,75 7,1 7 6 5,16 5 4,7 German learners Spanish learners 4 3 2,49 1,8 2 1 0 how ever therefore nevertheless Figure 5. Comparison of the use of the three connectors in ILa and Ilb Interestingly, with the exception of nevertheless, the use of the connectors by German learners was more similar to that of the Spanish learners than to the use of the correspondents of these connectors in native German. According to the DEE model, in which such similarities between the interlanguages of learners from different language backgrounds point to the unlikelihood of transfer, this indicates that transfer from German does not have a significant impact on the use of connectors by German learners. Rather, their behavior in that respect, be it correct or not, is in-line with what other learners with the same level of proficiency in English do. The two comparisons at the first stage of the DEE model, hence, lead to the assumption that transfer from German is not a major influence on the quantitative connector usage by German learners of English. The subsequent explanation stage (Step 3), where the model intends to trace possible reasons for transfer, should bring clarity to the matter. It aims to determine whether English and German are two languages between which transfer is likely to occur with regards to connector usage. For this purpose, the DEE model establishes the 'language distance' between the native and the target language. When comparing the frequencies of the connectors in question in native German and native English, divergent values can be observed both when only considering connectors a well as when taking all functional correspondents into account (Figure 6). Connector Usage in Essays Written by German Learners of English 27 Figure 6. Comparison of the use of the three connectors in TL and NL The considerable deviation in the frequencies of however and its correspondents once more points to the necessity of factoring in but and aber. They occur at similar rates in the English and the German texts, namely 42.41 versus 38.67 times per 10,000 words. Apparently, contrast is expressed at similar frequencies in German and English, but it is not done by using a connector in German (jedoch, allerdings) as much as it is in English (however). Therefore, in contrast, occurred much less frequently in the native English texts than its functional correspondents in native German. This supports the observation made earlier, that therefore is only one possible item to express a resultive relation but it may also indicate that consequence is left implicit more frequently in English than in German. This difference makes transfer implausible and rather implies that the learners need to be introduced to the different discourse conventions of German and English The third connector, nevertheless, also occurred only about half as much in native English as did its correspondents in German. This might again point do different discourse conventions. In sum, English and German seem to be working differently with regards to the frequency of the connectors under investigation and the discursive functions they serve. The distance between them is obvious in this respect. This suggests, according to the DEE model, the unlikelihood of transfer and may be referred to as a possible argument against the phenomenon. In order to evaluate any possible instances of transfer, in a last step the frequencies of the connectors in non-native English (IL) and native English (TL) are compared (Step 4). As can be seen in Figure 9, the figures were strikingly analogous. Connector Usage in Essays Written by German Learners of English 10 8,75 28 8,35 8 5,16 6 6,17 4 IL 2,49 2 TL 1,8 0 however therefore nevertheless Figure 7. Comparison of the use of the three connectors in IL and TL The German learners' interlanguage was closer to the native target than to their mother tongue. For all three connectors, the frequencies in the essays written by German learners of English were closer to those in native English than to the correspondents in German. This also seems to disprove the influence of transfer. Figure 8 provides an overview of the results that the four comparisons of the DEE model yielded. Figure 8. Overview of the use of the three connectors in the four varieties The NL/IL analysis (Step 1) revealed few similarities in the usage of connectors in German learners' interlanguage and native German. In nearly all cases the learners' use of connectors was very different from the use of the correspondents in their mother tongue. Therefore, L1 interference is unlikely to have occurred. The comparison of the interlanguage of German learners with that of Spanish learners (Ila/ILb, Step 2) furthermore showed that the latter use the connectors in question to similar extents. Hence, definite instances of language-specific transfer from German to the interlanguage have not been detected. In the next step (Step 3), the fact that the language distance between German and English with regard to the frequencies of however and therefore is quite considerable added to the impression that transfer Connector Usage in Essays Written by German Learners of English 29 does not play a major role in the connector usage of German learners of English. Finally, the comparison of the frequencies of however, therefore and nevertheless in IL and TL (Step 4) yielded almost analogue values. Negative transfer can therefore be excluded from possible factors affecting the quantitative usage of the three connectors under investigation. The problems they do seem to concern functionality rather than frequency. This implicates that learners have some connectors in their English lexicon, but they lack the ability to use them correctly, which cannot be explained by L1 transfer, either This confirms the hypothesis formulated at the outset of this study. 3.3 Analysis of under- and overused connectors In order to find out what, if not transfer, is the cause of the problems for German learners of English, it is necessary to find more connectors they have difficulties with. This is why it is worth looking at some more connectors that stood out when first inquiring the most frequent connectors in the text samples. Some connectors that were significantly underused by the German learners could be noticed. Figure 9 displays the four connectors underused by German learners with their frequencies in native English (TL)learner English (IL) and native German (NL). 11,4 12 10 8 8,16 6,64 TL 4,93 6 3,32 4 2 0,65 1,931,8 IL 4,46 2,3 2,582,28 NL 0 yet for example thus though Figure 9. Underused items in the learners’ texts compared to the TL The most striking case of underuse is the contrastive connector yet. Native speakers used it an impressive 6.64 times per 10,000 words. The German learners, by contrast, used yet as a contrastive connector no more than 0.65 times per 10,000 words. Like that of nevertheless, the contrastive function of yet would be realized in German as trotzdem or dennoch, which occurred 3.32 times per 10,000 words in the native German sample. Transfer, thus, cannot be the reason for the underrepresentation of Connector Usage in Essays Written by German Learners of English 30 yet in German learners' interlanguage. Rather, it might be due to their sheer ignorance of that use of the word. Contrastive yet is a formal alternative to the much more common and familiar however, and learners might simply not be aware of that. This assumption is supported by the fact that Leńko-Szymańska (2008), who studied the connector usage by learners from seven different mother tongue backgrounds, and Altenberg&Tapper (2008), who did the same for Swedish learners, also observed an underuse of yet in all the learner texts (cf. Leńko-Szymańska 2008: 103, Altenberg&Tapper 2008: 87). This might point to inadequate teaching of the use of connectors in Europe in general. Here, then, is first proof for the hypothesis put forward earlier, that the teaching of connectors in the EFL classroom is inadequate. The use of the appositive connector for example and its abbreviation e.g. is similarly diverging in the texts by German learners and by native speakers of English. Native speakers used it 4.93 times per 10,000 words, the learners only 1.93 times. The learners' use of this connector is more similar to that in their native language: In the native German texts, zum Beispiel/z.B. occurred 1.8 times per 10,000 words. This suggests the influence of negative transfer, because in formal German, examples are often paraphrased and not introduced directly. In English, on the other hand, connectors such as for example or also for instance are acceptable and common even in formal texts. Quirk et al. include them in their lists without any annotation that would indicate their belonging to more informal genres (cf. Quirk et al. 1985: 636). Learners might simply be unaware of this difference in discourse conventions and therefore rely too much on their mother tongue. This, then, would again not so much be a problem of transfer but of inadequate teaching. Transfer of German conventions can definitely be said to take place here. However, it is not the cause but the result of the problem, which really is that learners were not taught how to use for example correctly. The third underused connector is the summative thus (used 4.46 times per 10,000 words by native speakers, 2.3 times by German learners). Thus can be used as a formal alternative to therefore (cf. Quirk et. al.: 635), which is the second mostused connector in the sample of German learner writing. This implies that the learners do express results and consequences in their English essays, they are aware of that function. They simply do not know the alternative thus and that it is more appropriate that therefore in formal genres. The functional German equivalents daher, deshalb, also, folglich, demzufolge, somit and mithin occur 11.40 times per Connector Usage in Essays Written by German Learners of English 31 10,000 words in the native German sample. The strikingly low representation of thus in German learners' interlanguage compared to its commonness in native English and the high frequency of the resultive function in native German rules out the likelihood of transfer. Instead, I would once more suspect learners' general ignorance of alternatives because the information they received and the materials they were provided with were inadequate. A fourth case of underuse that needs to be mentioned is though. As pointed out earlier, though was also amongst the connectors used most frequently by the German learners, albeit it was not always used as a connector in Quirk et. al.'s sense. To them, though is only a connector or conjunct when it is used as a concessive contrastive, that is, when connecting two units of which “one unit is seen as unexpected in the light of the other” (ibid.: 639). But even if all instances of a contrast expressed by though by German learners, both as a connector and as a subordinator, would have been counted (2.49 occurrences per 10,000 words), it is used far more often by native speakers of English, namely 8.16 times per 10,000 words. Again, transfer from German cannot account for this difference. In native German, the equivalent obwohl, which is used 2.28 times per 10,000 words in the sample, cannot serve as a reliable reference for the use of though in English because it corresponds to sentence-initial though, only. Where though occurs in the middle or at the end of a sentence, however, it could be translated as, for example, allerdings,and jedoch. But even if we consider all functional correspondents expressing the contrastive function of though in German (obowhl, allerdings, jedoch and trotzdem), they still occur less frequently (5.89 times per 10,000 words) than though does in native English. Transfer from German cannot account for the low representation of though, since it occurs even less frequently in the learners' interlanguage than in native German, and is underused by Altenberg & Tapper's Swedish subjects as well (Altenberg&Tapper 2008: 87). Apparently many learners are not aware of the appropriate usage of though, and this might once more be due to teaching-induced factors. The analysis of overused connectors leads to similar findings. Figure 10 shows the two connectors which the German learners used far more often than the native English speakers. Connector Usage in Essays Written by German Learners of English 3 32 2,56 2,5 2 TL IL NL 1,5 0,82 1 0,5 0 0 0,57 0,09 by the way 0,09 anyway Figure 10. Overused items in the learners’ texts compared to the TL The most striking case of overuse was the connector anyway. The German learners use it 2.56 times per 10,000 words, whereas the native speakers use it only 0.09 times. It is implausible that this overuse is a result of transfer, as the most probable German correspondents, sowieso and ohnehin, also occurred only 0.57 times per 10,000 words. The problem seems to stem from learners' unawareness of the registerappropriateness of the connector. Anyway is listed in Quirk et al.'s list of connectors with the annotation <informal>(cf. Quirk et. al.: 635). Since German learners are often exposed to informal English in their private lives, in popular music, movies, television shows and the internet, for instance, they may confuse registers. Registerunawareness as a cause for misuse is also suggested by Altenberg & Tapper (2008: 87, 92). It is the task of teachers to counter this by raising student's awareness of different registers and genres and providing them with alternatives for the respective styles. This task was apparently not performed sufficiently in the case of the German learners. The second overused connector is by the way. The frequency in the learner texts, 0.82 times per 10,000, is not very high but is nevertheless noteworthy, because in the native English texts by the way does not occur at all. The German equivalent übrigens is not used frequently, either: only 0.09 times per 10,000 words. Again, learners cannot have been influenced by their mother tongue when they overused by the way. Instead, this seems to be yet another case of teaching-induced misuse. According to Quirk et al, “transitional conjuncts” like by the way “serve to shift attention to another topic or to a temporally related event” (Quirk et al.:636). The very point of argumentative writing, however, is usually to concentrate on one topic and not deviate from it. Professional writers avoid side notes and digressions. They are more of a characteristic of spontaneous talk. If the learners had been made aware Connector Usage in Essays Written by German Learners of English 33 of this stylistic distinction, they would not use by the way as often in their written production. The analysis of those connectors which proved problematic for German learners of English showed that difficulties cannot be attributed to transfer from their mother tongue. Instead, a general unawareness of register-appropriateness and discourse conventions and an ignorance of alternative connectors for distinct registers can be assumed. This assumption is in line with observations made in previous studies on the topic (cf. Leńko-Szymańska 2008, Altenberg&Tapper 2008, Gilquin&Paquot 2008). Inadequate teaching materials and inappropriate methods to convey the correct use of connectors seem to have a greater impact on the written production of German learners of English than transfer from their mother tongue. Didactic approaches therefore urgently require revision in that respect. The following section will investigate to what extent the results of this study can be used to draw conclusions for EFL teaching and inform the development of more adequate teaching methods and materials. Connector Usage in Essays Written by German Learners of English 34 4 Pedagogical implications and outlook The analysis of the connector usage of German learners of English has revealed that, to all appearances, L1 interference is not a major cause for the problems learners seem to have with this discursive device. Rather than misusing individual connectors because of a negative influence of their mother tongue, the learners seemed to have only a limited repertoire of connectors at their disposal. This repertoire does not suffice to meet the requirements of all genres, in the present case academic writing. Similar observations have already been made by Altenberg&Tapper (2008), LeńkoSzymańska (2008), and Gilquin&Paquot (2008), who all note that learners seem to lack knowledge of register-appropriate alternatives to the few connectors they frequently use. What is interesting is that these studies worked with corpora of texts produced by learners from native language backgrounds other than German and nevertheless achieved analogous results. This suggests that the problems with connector usage are not restricted to German learners, but are a general problem for EFL students. If L1 transfer as a possible cause for problems can be excluded, other explanations have to be found. The studies quoted above concordantly offer teaching-induced factors as one source of problems. They criticize that not enough attention is paid to familiarizing students with different registers and training them in academic writing in the EFL classroom (cf. Altenberg&Tapper 2008: 92). Learners apparently have not been adequately taught how to create coherence in their texts (cf. Leńko-Szymańska 2008: 104). There is ground for the assumption that the teaching of connectors in the EFL classroom in Europe (where all of the subjects in the studies quoted above come from) is inadequate. Both Gilquin & Paquot (2008) and Milton & Tsang (1993) adduce evidence for this from EFL textbooks. The respective excerpts are reproduced in Figures 11 and 12. Figure 11. Connectives of result in Jordan (1999: 62) as reproduced in Gilquin&Paquot (2008) Connector Usage in Essays Written by German Learners of English 35 Figure 12. Use of discourse markers in Burns&Smallwood (1990: 110) as reproduced in Milton&Tsang (1993: 232) Gilquin&Paquot argue that such “undifferentiated lists of connectors” as the one in Figure 13 have a “pernicious influence” on learner output. (Gilquin&Paquot 2008: 54). They give the misleading impression that the connectors were synonymous, interchangeable, and equally valid in different genres and registers. Similar criticism is expressed by Milton & Tsang, who find fault with lists like the one in Figure 14 because they ignore syntactic and semantic differences and do not provide enough examples (cf. Milton&Tsang 1993: 17). Furthermore, they, too, criticize that such lists suggest that a given connector could simply be replaced with any other from the same category (cf. ibid.: 18). In a different paper, Milton sums up the problem as follows: Students are drilled in the categorical use of a short list of expressions – often those functioning as connectives or alternatively those which are colorful and complicated (and therefore impressive) – regardless of whether they are used primarily in spoken or written language (if needed at all), or to which text types they are appropriate. (Milton 1998: 190). The conclusion to be drawn from this criticism is that connectors are evidently imparted in an insufficient and misleading way in many cases. However, this does not mean that connectors are given too little attention in the EFL classroom. On the contrary, Leńko-Szymańska (2008) claims that connectors are often presented as a kind of panacea to improve learners' writing style and are hence treated extensively in class (cf. Leńko-Szymańska 2008: 104). Learners are therefore made believe that Connector Usage in Essays Written by German Learners of English 36 “if their essays contain a large variety and number of connectors, they will receive higher marks, which can lead to the forced and unnatural use of linking expressions” (ibid: 105). The often redundant use of however, therefore, and nevertheless by some of the German learners observed in Section 3 backs up this assumption. Learners seem to feel obliged to use connectors in their writing, while indeed forgoing them would in many cases make their writing more professional and native-like. LeńkoSzymańska (2008) compared the students' essays to professional writing in her study and found that the experts used significantly less connectors. She conjectures “that professional writers achieve coherence in their texts without an abundant use of linking expressions. […] [T]hey choose and structure their arguments more effectively than do inexperienced writers; thus the reader does not need many overt markers to follow the reasoning of the writer” (ibid: 104). The “abundant use of connectors”, then, seems to be “a general characteristic of novice writing” (ibid.) that is enforced by misleading teaching materials. There seems to be no doubt among learner corpus researchers that EFL teaching as it is practiced in many places at present is not in accordance with what corpus linguistics revealed about actual language use (both by native and by nonnative speakers) and problem areas (cf. Kaszubski 1998: 175, Römer 2006: 126). Thus, nearly all research papers in learner corpus research end with a section devoted to 'pedagogical implications'. Based on their respective findings, the authors pass criticism on current teaching methods and formulate a multitude of proposals to improve the imparting of the linguistic phenomenon under investigation. However, these well-intentioned proposals rarely make it into actual teaching practice. The supposed “direct link between uncovering differences between learner and native data and designing remedial pedagogical tools and methods” (Granger 2009: 22) insinuated in such proposals does not exist. Corpus linguists acknowledge that their findings “have not yet exerted such a strong influence on EFL textbooks” (Römer 2009: 90). A survey conducted by Mukherjee (2004) amongst practicing English teachers in Germany revealed that 79,4% of them did “not know anything about corpus linguistics” (Mukherjee 2004: 241f). He concludes that there is, at present, a large gap between the wealth of applied corpuslinguistic research and the teaching practice in Germany, which so far has only been affected to a very limited extent by this research. Closing this gap is a challenge to applied corpus linguists and, perhaps more importantly, to those who are involved in teacher training (ibid.: 247). Connector Usage in Essays Written by German Learners of English 37 Recently, corpus linguists have started to put the general call for a corpus-driven revision of EFL teaching into practice by designing concrete methods and materials. The general trend is to move from the previous “Present – Practise- Produce paradigm” to the “Observe – Hypothesise – Experiment paradigm” (Granger 2007: 62). This is achieved by the implementation of data-driven learning (DDL). This method uses authentic NS and NNS corpus data to familiarize learners with the differences between the respective varieties, common uses of specific items and typical mistakes in a ‘hands-on’ manner. One example of this approach as adapted in the Macmillan English Dictionary for Advanced Learners (Rundell 2007) can be seen in Figure 13. By drawing attention to the respective frequencies in native and non-native writing, the learners’ awareness of inappropriate overuse is raised (cf. De Cock & Paquot 2009: 200). Figure 13. Comparison of quantifiers in NS and NNS writing in Rundell (2007) as reproduced in De Cock&Paquot (2009: 200) After already having made their way into the design of dictionaries and grammars, corpus-linguistic findings are very slowly starting to be implemented in the creation of teaching materials and exercises, as well. ‘Fill-in-the-gap’ exercises based on real concordance lines generated by corpus-linguistic software (see Figure 14) are one example of how to familiarize learners with authentic language use. Connector Usage in Essays Written by German Learners of English 38 Figure 14. Example for a corpus-based DDL exercise as reproduced in Römer (2006: 125) The use of connectors is one area where such exercises could be perfectly suitable. Whereas traditional textbooks have learners practice the use of connectors with unnatural and decontextualized examples (cf Römer 2009: 90), exercises based on language as it is actually used can be expected to lead to a more profound understanding of which connector is appropriate in which context. As to the repertoire of connectors to chose from, the proposal formulated by Kaszubski (1998) for the enhancement of materials for Polish learners of English is also applicable to the case of connector usage by German learners. Future teaching materials about connectors should include more exhaustive “lists of synonymous items, accompanied with frequency band information, register/style description, and (gradable) overuse/underuse/misuse warnings”, and supplemented with NL/TL contrasting samples (Kaszubski 1998: 183). The main challenge, however, is not equipping the learners with the appropriate connectors, but teaching them when not to use them and how to create cohesion in their essays without them. Leńko-Szymańska's proposal is to continue teaching connectors, but to de-emphasize their importance in the construction of coherent discourse and focus on alternative ways of creating coherence (Leńko-Szymańska 2008: 106) She admits that “[s]uch a skill might be much more difficult to teach, but as a result, essays written by English learners may be not only more native-like, but also more expert-like” (cf. ibid.). The shift of focus in learner corpus research from lexical to more discursive phenomena just dispraised in the introduction of this paper comes now just at the right time, as the requested revision of teaching cohesion in EFL writing needs to be based on the findings of a corpus-based discourse analysis rather than a quantitative inquiry of connectors in learner writing. Connector Usage in Essays Written by German Learners of English 39 This study has shown that the analysis of German learners’ interlanguage and the comparison of it with their mother tongue, their target language and the interlanguage of learners with different native languages provides valuable insights into the problems and insecurities the learners have with respect to the use of connectors in their English essays. These problems could not, in contrast to what is frequently claimed in literature, be attributed to transfer from the mother tongue but to the inadequate teaching of connector usage in the EFL classroom. The findings of the corpus-linguistic analysis did not only help to detect this problem, they can also be used to find a remedy. For one thing, samples and excerpts from the corpus data or even the corpora themselves can physically be brought into the classroom (cf. Römer 2006: 124). In addition, the findings can be used to revise current teaching materials, considering the newly-gained insights about over-, under-, and misuse of specific connectors as well as the exaggerated frequency of connectors in learner texts in general, factors which had hitherto been disregarded in the majority of textbooks (cf. ibid.: 177). However, corpus linguistics can only make these valuable contributions to SLA and ELT research if studies like the present one continue to clear up prevalent misconceptions about learner language and the learning process that are reflected in many of the teaching materials on the market. For this purpose, more studies that combine a qualitative and a quantitative analysis of learner language are needed. The scope of this study was admittedly limited. Future studies should work with larger corpora and analyze more than three connectors. Working with parallel corpora that contain translations of the same texts would allow for a more accurate comparison of source and target language (cf. Gilquin 2008: 15). Similarly, the use of corpora containing texts of different genres and by people with different proficiency levels and expertise would lead to more comprehensive results (cf. Hasselgård/Johansson: 56). Furthermore, the analysis of a longitudinal corpus would be a great interest for the investigation of interlanguage. Such an analysis would offer substantial insights into the development of learner language over time and under specific circumstances (cf. ibid). In any case, the realization of such studies requires the availability of resources. Many of the necessary corpora are only commercially available or even need to be specially compiled for specific research purposes, which is an extremely expensive and time-consuming endeavor. Such projects will thus only be supported and funded if their usefulness is acknowledged by the respective authorities of education. It is necessary to convince practitioners Connector Usage in Essays Written by German Learners of English 40 and policy makers of how much they can benefit from learner corpus research. Römer (2006) suggests that this is best achieved by “foster[ing] communication among and improv[ing] the exchange between researchers on the one side and practitioners (teachers, teacher trainers and trainees, and materials writers) on the other side, so that more people see that corpora are immensely valuable tools in a language learning and teaching context” (Römer 2006: 129). Extensive publishing and fruitful international exchange has taken place amongst corpus linguists in the last twenty years, but outside of their circle their research has not attracted much interest. Yet, the transfer of academically achieved research findings onto real life problems and purposes is what the very discipline of applied linguistics is in essence about. Studying language for its own sake undoubtedly has its justification, but in effect language, and a lingua franca like English even more so, is usually studied in order to be taught, learned, and, most importantly, used in the global environment. I hope that with my bachelor thesis I could make a small contribution to “bridging the gap between applied corpus linguistics and the reality of ELT in Germany” (Mukherjee 2004) and demonstrate one possible way of bringing intention and implementation in applied linguistic research a little bit closer to one another. Connector Usage in Essays Written by German Learners of English 41 Works cited Altenberg, Bengt and Tapper, Marie (1998): “The Use of Adverbial Connectors in Advanced Swedish Learners' Written English”. In: Granger, ed. (1998): pp. 80 – 93. Burnard, Lou (2009): What is the BNC?. Official Website of the British National Corpus. University of Oxford. Accessible at: http://www.natcorp.ox.ac.uk /corpus/index.xml. Last Access: 10.06.2013. De Cock, Sylvie and Paquot, Magali (2009): “The Monolingual Learners' Dictionary as a Productive Tool: The Contribution of Learner Corpora”. In: Campoy, M. C., Bellés-Fortuno B., and Maria Luisa Gea-Valor, eds. (2009): CorpusBased Approaches to English Language Teaching, London & New York: Continuum. pp 195-204. Evenson, Lars S. and Rygh, Irmgard L. (1988): “Connecting L1 and FL in Discourse-level Performance Analysis.” In: Papers and studies in Contrastive Linguistics, 22. pp. 133-178. Gilquin, Gaëtanelle (2008): “Combining Contrastive and Interlanguage Analysis to Apprehend Transfer: Detection, Explanation, Evaluation”. In: Gilquin G., Papp S. & Díez-Bedmar M.B., eds. (2002): Linking up Contrastive and Learner Corpus Research. Amsterdam & Atlanta: Rodopi. pp. 3-33. Gilquin Gaëtanelle, and Paquot Magali (2008): “Too Chatty: Learner Academic Writing and Register Variation”. In: English Text Construction 1 (1). pp. 4161. Granger, Sylviane (2002): “A Bird's-Eye View of Computer Learner Corpus Research”. In: Granger S., Hung J. and Petch-Tyson S., eds. (2002): Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching. Amsterdam & Philadelphia: Benjamins. pp. 3-33. Granger, Sylviane, ed. (1998): Learner English on Computer. London & New York, Addison Wesley Longman. Granger, Sylviane (1998): “The Computerized Learner Corpus: A Versatile New Source of Data for SLA research”. In: Granger, ed. (1998): pp. 3 – 18. Granger Sylviane (1996): “From CA to CIA and Back: An Integrated Approach to Computerized Bilingual and Learner Corpora”. In: Aijmer K., Altenberg B. and Johansson M., eds. (1996): Languages in Contrast. Text-based Crosslinguistic Studies. Lund: Lund University Press. pp. 37-51. Granger ,Sylviane and Tyson, Stephanie (1996): “Connector Usage in the English Essay Writing of Native and Non-native EFL Speakers of English". In: World Englishes, 15. pp. 19-29. Hasselgård, Hilde and Johansson, Stig (2011): “Learner Corpora and Contrastive Interlanguage Analysis”. In: Meunier, Fanny, Sylvie De Cock, Gaëtanelle Connector Usage in Essays Written by German Learners of English 42 Gilquin and Magali Paquot, eds. (2011): A Taste for Corpora: In Honour of Sylviane Granger. Amsterdam: Benjamins. pp. 33–62. Hasselgren, Angela (1994): “Lexical Teddy Bears and Advanced Learners: A Study into the Ways Norwegian Students Cope with English Vocabulary”. In: International Journal of Applied Linguistics 4 (2). pp. 237-258. Herriman, Jennifer and Böstrom Aronsson, Mia (2009): “Themes in Swedish Advanced Learners' Writing in English”. In: Aijmer, Karin, ed. (2009): Corpora and Language Teaching. Amsterdam: Benjamins. pp 101–120. Institut für Deutsche Sprache (2002): Das Deutsche Referenzkorpus – DeReKo. Official DeReKo Website. Institut für Deutsche Sprache, Mannheim. Accessible at: http://www.ids-mannheim.de/allgemein/impressum.html. Last Access: 10.06.2013. Jarvis, S. (2000): “Methodological Rigor in the Study of Transfer: Identifying L1 Influence in the Interlanguage Lexicon”. In: Language Learning, 50(2). pp. 245-309. Jordan, R.R. (1999): Academic Writing Course. Study Skills in English. Third Edition. Harlow: Longman. Kaszubski, Przemysław (1998): “Enhancing a Writing Textbook: A National Perspective”. In: Granger, Sylviane, ed. (1998). pp.172-185. Kohn, Kurt (1986): “The Analysis of Transfer”. In: Kellerman, E., and Sharwood Smith, M., eds. (1986): Crosslinguistic Influence in Second Language Acquisition. Oxford: Pergamon Press. pp. 21 – 34. Leńko-Szymańska, Agnieszka (2008): “Non-native or Non-expert? The Use of Connectors in Native and Foreign Language Learners’ Texts”. In: Acquisition et Interaction en Langue Etrangère 27. pp. 91-108. Milton, J. and Tsang, E.S.C. (1993): “A Corpus-Based Study of Logical Connectors in EFL Student's Writing: Directions for Future Research. In: Pemberton, R. and Tsang, E.S.C., eds. (1993): Lexis in Studies. Hong Kong: Hong Kong University Press. pp. 215-246. Milton, J. (1998): “Exploiting L1 and Interlanguage Corpora in the Design of an Electronic Language Learning and Production Environment”. In: Granger, Sylviane, ed. (1998). pp. 186-198. Mukherjee, Joybrato (2004): “Bridging the Gap between Applied Corpus Linguistics and the Reality of English Language Teaching in Germany”. In: Connor, Ulla and Upton, Thomas A., eds. (2004): Applied Corpus Linguistics. A Multidimensional Perspective. Amsterdam: Rodolpi. pp. 239-250. Nesselhauf, Nadja (2003): “The Use of Collocations by Advanced Learners of English and Some Implications for Teaching”. In: Applied Linguistics 24 (2). pp. 223-242. Connector Usage in Essays Written by German Learners of English 43 Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, Jan Svartvik (1985): A Comprehensive Grammar of the English Language. London/New York: Longman. Römer, Ute (2006): “Pedagogical Applications of Corpora: Some Reflections on the Current Scope and a Wish List for Future Developments.” In: Zeitschrift für Anglistik und Amerikanistik 54 (2). pp. 121-134. Römer, Ute (2009) “Corpus Research and Practice: What Help Do Teachers Need and What Can We Offer?” In: Aijmer, Karin, ed. (2009): Corpora and Language Teaching. Amsterdam: Benjamins. pp. 83-98. Rundell, M., ed. (2007): Macmillan English Dictionary for Advanced Learners (second edition). Oxford: Macmillan Education. Selinker, Larry (1972): “Interlanguage”. In: International Review of Applied Linguistics, 10. pp. 209-241. Scholze-Stubenrecht, Werner, ed. (2008): The Oxford German Dictionary. Third Edition. Oxford: Oxford University Press. Scholze-Stubenrecht, Werner, ed./Bibliographisches Institut (2011): Duden. Deutsches Universalwörterbuch. Seventh Edition. Mannheim: Dudenverlag 2011. Connector Usage in Essays Written by German Learners of English 44 Erklärung zur Urheberschaft Hiermit versichere ich, dass diese Arbeit von mir persönlich verfasst wurde und dass ich keinerlei fremde Hilfe in Anspruch genommen habe. Ebenso versichere ich, dass diese Arbeit oder Teile daraus weder von mir selbst noch von anderen als Leistungsnachweis andernorts eingereicht wurden. Wörtliche oder sinngemäße Übernahmen aus anderen Schriften und Veröffentlichungen in gedruckter oder elektronischer Form sind gekennzeichnet. Sämtliche Sekundärliteratur und sonstige Quellen sind nachgewiesen und in der Bibliographie aufgeführt. Das Gleiche gilt für graphische Darstellungen und Bilder sowie für alle Internet-Quellen. Cathrin Hein-Becker Gießen, den
© Copyright 2026 Paperzz