Sie sind auf Seite 1von 26

Contrast and Concession: The Use of However, Nevertheless, Yet and Still in Native and NonNative Student Writing

Brian Davies

Department of English Bachelor Degree Project English Linguistics Autumn 2010 Supervisor: Annelie del

Contrast and Concession: The Use of However, Nevertheless, Yet and Still in Native and Non-Native Student Writing
Brian Davies

Abstract
Previous comparative studies of usage of contrast markers have found that Swedish non-native speakers of English underuse and overuse different individual items compared to native speakers. This paper compares the use of the contrast markers however, nevertheless/nonetheless, yet and still in essays in linguistics to determine how and why they might be used differently. Essays are taken from the Stockholm University Student English Corpus and are equally represented by examples from Kings College and Stockholm University. Random samples are analyzed for type of contrast signalled and repeated collocational patterns. The findings show that the Swedish students underuse however and overuse nevertheless/nonetheless, yet and still in signalling concessive relationships. The research suggests that the Swedish students have a weaker grasp of prototypical frequency of usage than native speakers, which may result from native language transfer. It is argued that it would be beneficial for Swedish learners of English, and by extension other L2 learners, to improve awareness of constructing effective contrastive relationships.
Keywords

Academic writing, contrast markers, collocational patterns, concessive relationships, prototypical usage, native language transfer

Contents
1. Introduction 1

2. Background ................................................................................... 2
2.1 Concession .........................................................................................2 2.2 Non-concession ...................................................................................4 2.3 The prototypical semantic senses of contrast markers ..............................5

3. Method ......................................................................................... 5
3.1 Stockholm University Student English Corpus ..........................................5 3.2 Qualitative analysis of contrast markers .................................................7
3.2.1 Discarding texts not signalling contrastive relationships ................................. 7 3.2.2 Analyzing the texts at the ideational level for concession................................ 8

3.3 Co-occurrence ................................................................................... 10

4. Results ....................................................................................... 10
4.1 Use of however as a concessive and non-concessive marker ................... 10 4.2 Use of nevertheless/nonetheless, still and yet as concessive and nonconcessive markers ................................................................................. 12 4.3 However and Nevertheless in the same paragraph ................................. 13 4.4 Collocations & Hedging ....................................................................... 14 4.5 Concession within the academic essay .................................................. 16

5. Discussion ................................................................................... 16
5.1 Stylistic Uncertainty ........................................................................... 17 5.2 Native Language transfer .................................................................... 18

6. Conclusions ................................................................................. 19 References ...................................................................................... 21

1. Introduction
Writing an academic essay is like driving around the roads of a town or city. The author, instead of driving cars around, is driving arguments around. The arguments change direction, come to a halt or accelerate forwards to their destination. Like a driver uses the indicators to signal changes in the cars direction, a writer uses contrast markers to change the direction of the argument or take the argument logically forward (Granger & Tyson 1996). Driving abroad, we are sometimes taken aback by how the natives indicate and signal differently from our own culture. Likewise, different nationalities signal contrastive relationships in different ways. Research has shown that Swedish non-native speakers of English (NNS) underuse contrastive markers in academic essays compared to British native speakers (NS) (Altenberg & Tapper 1998; Heino 2010). Although however is the most frequently used contrast marker in academic essays by both NS and Swedish NNS, Swedish NNS have still been found to underuse however in comparison to NS (Altenberg & Tapper 1998; Heino 2010). At the same time, Swedish NNS have been found to overuse individual contrast markers; Altenberg and Tapper (1998) found overuse of still and Heino (2010) found overuse of still and nevertheless. There is a connection between however, nevertheless and still, which is that they all share the semantic sense of concession (Bell 2010). Evidence of the underuse of the concession marker however at the same time as overuse of the concession markers nevertheless and still suggests that an element of substitution may be taking place. In other words, Swedish NNS may be overusing nevertheless and still by using them in the place of however as a concession marker. This study attempts to show how British NS and Swedish NNS use contrast markers differently when writing in English. The contrast markers analyzed are however, nevertheless/nonetheless, still and yet. However, nevertheless and still have been chosen as they have been found to be used differently by NS and NNS in previous studies (Altenberg & Tapper 1998; Heino 2010). Nonetheless has been chosen because it is practically a synonym of nevertheless (Ball 2010). Altenberg (2002) studied English and Swedish concession markers in a bidirectional corpus study. The study examined how English concession markers were translated into Swedish and vice versa. The final contrast marker to be analyzed is yet, following Altenbergs findings that yet appears in the same concession set as nevertheless and still. This study uses as its primary source undergraduate academic essays by British NS and Swedish NNS from the Stockholm University Student English Corpus (SUSEC). It is a qualitative examination of the uses of contrast markers. The study is exploratory, as the contrast markers chosen are far from exhaustive. Moreover, there has been no comparative analysis of the ways in which NS and NNS construct contrastive relationships, especially concessive ones. Nevertheless, despite its limitations, this study attempts to partially determine what a specific group of Swedish NNS do in their writing instead of using however and the reasons for doing that.

The following research questions have been studied. Do a group of NS use however for concession more frequently than the studied Swedish NNS? Do the Swedish NNS use nevertheless/nonetheless, still and yet for concession more frequently than the NS? Are there further patterns of contrast marker usage that are evident in the NS but absent in the Swedish NNS? If differerences are found, what may the possible causes be?

2. Background
However is an example of a discourse marker. A discourse marker is ordinarily a conjunction, adverbial or prepositional phrase that signals that a prior unit of discourse, S1, has a relationship with the following unit of discourse, S2, which contains the discourse marker (Fraser 1998). In this paper I have used the term contrast marker, to describe discourse markers that signal a relationship of contrast between S1 and S2. In example (1) below She didn't get the award after all is S1, Still, her results were very good is S2, in which still is the contrast marker. Contrast can be seen as comparing the ideas in S1 and S2 and at the same time showing the differences. Linguistically, the notion of contrast is considerably more complex as it contains a number of contrasting (!) categories. It is not easy to find agreement on the different categories of contrast amongst linguistics (Quirk et al. 1985: 635-636; Fraser 1998; Mann & Thompson 1988). Different categories of contrast have variously been labelled reformulatory, replacive, antithetic, concessive, corrective, true/false and antithetical contrast. The aim of this study is not to define contrast and its component categories. However, this study does require a usable definition of concession as naturally occurring discourse is taken from the SUSEC corpus and contrastive relationships are described as concessive or non-concessive. The following section outlines the definition of concession applied in analysing the naturally occurring discourse in this study.
2.1 Concession

Quirk et al. (1985) describe concession as a situation where one unit of discourse is unexpected in the light of the other, as seen in example (1). (1) She didn't get the award after all. Still, her results were very good. (Quirk et al. 1985: 639)

Fraser (1998) describes the concessive unit of discourse (S2) as targeting an indirect message, or expected implication, in the previous unit of discourse (S1), as seen in example (2). (2) We started late. (Expected implication: We will arrive late) Nevertheless we arrived on time. (Fraser 1998: 318).

Although these definitions are helpful, they cannot always be applied to naturally occurring discourse in determining whether a contrastive relationship is concessive or non-concessive. Example (3), taken from the SUSEC corpus, shows a contrastive relationship. The emphasising use f italics is mine in this example as in the other examples of thr essays used as empirical material. (3) There are areas where there may be accusations of this language policy infringing linguistic human rights'; however, as mentioned in the introduction, the choice has been made by the immigrant to learn English. (KC 3 LING 30). Using Quirk et al.s definition it is hard to determine whether the relationship is concessive or non-concessive. According to Quirk et al., a concessive relationship would entail that the choice ... made by the immigrant to learn English was unexpected. From the context this is difficult to determine. Example (4) shows a contrastive relationship taken from SUSEC. (4) Paragraph 11 would initially seem to be a negative for Liverpool; however the quote from Benitez is that may be a positive. (KC 3 LING 26) According to Fraser, this would be an example of non-concessive contrast. The message in S1 is direct (initially seem to be a negative), and concessive relationships demand an indirect message is targetted. However, there is a sense of concession in this example. The author concedes that paragraph 11 would seem negative at first glance. Therefore a more usable definition of concession is needed. Mann & Thompson (1988) labelled contrasting ideas, S1 and S2, as satellite and nucleus, or satellite and satellite. If S1 and S2 have the labels satellite and satellite, they have equal importance as seen in example (5). (5) Animals heal but trees compartmentalize. They endure a lifetime of injury and infection by setting boundaries that resist the spread of invading microorganisms. (Mann & Thompson 1988) Where S1 is labelled satellite and S2 is labelled nucleus (or vice versa), the nucleus is more important as shown in examples (6) and (7). Mann and Thompson further divide satellite / nucleus relationships into antithesis and concession relationships. Concession can be defined by the two following qualities; the writer wishes the reader to have a higher regard for the nucleus, and the writer concedes that the satellite should not be

entirely discounted, as seen in example (6). Antithesis is differentiated from concession in that S1 and S2 are seen as incompatible as shown in example (7). (6) Although [dioxin] is toxic to certain animals, evidence is lacking that it has any serious long-term effect on human beings. (Mann & Thompson 1988)

(7) [T]he tragic and all too common tableaux of hundreds or even thousands of people snake-lining up for any task with a paycheck illustrates a lack of jobs not laziness. (Mann & Thompson 1988)

Mann & Thompsons definition of concession has been chosen as the easiest to apply to naturally occurring discourse. If we apply the Mann & Thompson definition to example (3), we can determine a concessive relationship. S1 describes a language policy infringing human rights. S2 describes a choice made by immigrants to learn English. S2 seems to have more importance as it includes a supporting phrase as mentioned in the introduction. Therefore, the writer concedes that accusations of infringing human rights might be valid, but more importantly the choice has been made freely. Of course, this is a subjective interpretation. In section 3.6 I will describe how I attempted to objectively apply the definition to naturally occurring discourse.
2.2 Non-concession

This study describes all forms of contrast other than concession, as non-concessive contrast. However, contrast is not necessarily superordinate of its hyponyms concession and non-concession. Instead, using Mann & Thompsons (1988) model (see Figure 1) contrast is an antonym of concession. Concession and antithesis are hyponyms of what Mann & Thompson term Antithesis/Concession. As detailed above, this study uses Mann & Thompsons definition of concession, but the Fraser (Bell) model (see Figure 1) is mentioned in the Discussion section.

Fraser (Bell) Classification General Contrast Correcting Contrast True/False Concession

Mann & Thompson Classification Relations between S1 and S2 Contrast Antithesis/Concession Antithesis Concession

(Adversative)

Figure 1. Alternative classifications of concession

Frasers (1988) classification describes concession as a hyponym of contrast, as shown in Figure 1. Fraser described six other sub-categories of contrast, which are not included in Figure 1. However, Altenberg (2002) combined two of the sub-categories

that Fraser had headed with the contrast markers however and on the other hand (Fraser 1988: 313 315) to create his adversative category of contrast. In the discussion of results, this paper makes reference to adversative contrast, which is based on the Fraser (Bell) classification.
2.3 The prototypical senses of contrast markers

Contrast markers have prototypical senses that are either concessive, non-concessive or both concessive and non-concessive (Fraser 1998). Nevertheless/nonetheless has a prototypical sense of concession, whereas however is rather more ambiguous and has senses which are both concessive and non-concessive. Yet and still are also prototypically concessive, but with subtle differences, with nevertheless/nonetheless being most concessive and yet least concessive, with still in the middle (Bell 2010). It is beyond the scope of this essay to test the strength of the concession in the examples used. Nevertheless, the use of contrast markers does have a semantic effect, and how these senses are applied to the naturally occurring discourse in the study is discussed in section 3.6 below.

3. Method
The research used a corpus of naturally occurring undergraduate linguistic essays which was then divided into two equally sized subcorpora of NS essays and Swedish NNS essays. The AntConc (Antony 2007) concordance software was used to count and show examples of uses of however, nevertheless/nonetheless, yet and still in both subcorpora. Paragraphs containing the studied contrast markers were analysed to decide whether the relationship described was concessive or non-concessive. Following this, paragraphs were examined to determine the extent of co-occurrence of however and nevertheless/nonetheless in the same paragraph. In addition, the software was used to search examples of recurring collocations including hedges.
3.1 The Stockholm University Student English Corpus

The Stockholm University Student English Corpus (SUSEC) is a large corpus of over one million words, containing 368 undergraduate English essays from Stockholm University and Kings College London. These essays are part of the course work of English undergraduate courses in terms one to four inclusive. Texts range from 500 to 14,000 words in length and include both linguistic and literature essays1.

For queries about the corpus, please contact Britt Erman at Stockholm University (Britt.Erman@english.su.se)

For this study the corpus was divided into two subcorpora: a NNS and a NS subcorpus. These two subcorpora were labelled respectively SU Ling and KC. The SU Ling subcorpus comprises linguistic essays from first and second term undergraduates in the English Department at Stockholm University. The KC subcorpus comprises linguistic essays from second and third term undergraduates in the Linguistics Depatment at Kings College, London. As both genre and discipline are likely to affect the quantity and usage of contrastive markers (e.g. Shaw 2009), both subcorpora reflect the same genre (academic essay) and discipline (linguistics). For this reason literature essays were not included. The same subcorpora were used in this study as the Heino (2010) quantitative research into adverbial connectors. Following Heinos findings of underuse of however and overuse of still and nevertheless by the Swedish NNS, this study intends to extend further her investigation by comparing use of however, nevertheless, still (and yet) by NS and NNS, using the same corpora. These subcorpora are well suited for a comparative study as they are relatively large, with SU LING containing 252,887 words and KC containing 247,425 (cf Altenberg & Tapper 1998), who compared NS and NNS corpora of both approximately 50,000 words). In addition the corpora are contemporary, from 2007. There are two criticisms that could be levelled at the choice of subcorpora. First, the subcorpora are not equally mature, as the KC subcorpus includes essays from students in their second and third terms of studies, whereas the SU LING corpus includes essays from students in their first and second term of studies. However, there are no examples of first term NS essays in SUSEC. Additionally, the third term Swedish NNS essays are BA dissertations and of a far more extensive nature than the essays included in the SU LING and KC subcorpora. A simple average word count illustrates this, as illustrated in Table 1. Therefore third term NNS essays were not included, as they are of the separate dissertation subgenre, rather than the essay genre.

Table 1. Average word count per text across subcorpora

NNS 1st term Average word 986 count Included subcorpus in SU LING

NS 2nd term

NNS 2nd term

NS 3rd term

NNS 3rd term

2,784 KC

2,111 SU LING

3,145 KC

6,738 No

Second, the fact that the KC subcorpus contains essays written in the UK does not entail that the writers were NS. Likewise, the essays written in Sweden could have come from NS. I suggest that this criticism is not decisive as this study is of an exploratory nature. However, if it is felt worthwhile to extend the study, as suggested in this essay, then it would be necessary to address this criticism by removing NS essays from the SU LING subcorpus and NNS essays from the KC subcorpus.

3.2 Qualitative analysis of contrast markers

The first stage of the qualitative analysis was to use the AntConc software to search the subcorpora for tokens of the studied contrast markers. Second, the texts were manually examined to determine whether the tokens were used as contrast markers or not, and examples of other uses were discarded. Finally, random samples of texts were manually analyzed to determine whether a concessive or non-concessive relationship was expressed.
3.2.1 Discarding texts not signalling contrastive relationships

However and nevertheless/nonetheless only function as discourse markers, with no other uses. Still and yet have functions other than signalling contrats, which meant each text returned by AntConc required analysing and, if necessary, discarding unwanted examples.

3.2.1.1 Determining still as a Contrast Marker

Still can have two functions within a text; it can function as a contrast marker or as a temporal adverb, where still as a temporal adverb indicates a continuing action or condition (Bell 2010). There is not always a clear division between the two functions, as shown in (8). (8) But they are still the works of the attesting spirit. (Bell 2010)

In example (8) we can substitute nevertheless for still and derive a contrastive relationship. Alternatively, still has the sense of continuing to have the condition, thus functioning as a temporal adverb. The important factor in determining the function of still is context. For each token of still I examined the context and determined the function. This is a subjective process, but there are indicators that help. Still as a temporal adjective, frequently appears in sentence medial position, immediately preceding or following a main or auxiliary verb. Substituting still for continue/s or continuing will keep the same meaning in the text. Still as a contrast marker frequently appears in clause initial position, sometimes collocates with but, and is often followed by a comma. Substituting still for nevertheless will keep the same meaning in the text. Examples (9), (10) and (11) show ambiguous examples from the SU LING subcorpus. Below I describe how I interpreted the function of still. (9) That confirm how they are affected by their immediate surroundings, because their parents are from Sweden, but they still talk like their friends. (SU A LING 01) (10) English and French coexisted for some centurys but at the fourteenth century the English language so to say won the war, maybe because it was simpler in grammar and richer in vocabulary. Still in this century people

from one part of Britain could not understand people from another part of the island. (SU A LING 64) (11) ...; but when describing a stranger over 35 years of age dam might be used even more willingly than kvinna. Still dam is not used without feeling a bit pretentious. As a result we have two well established words that are loaded with meaning for historical reasons. (SU A LING 06)

In example (9) we could substitute nevertheless/nonetheless for still and contrast an expectation that the actors would not talk like their friends, but in fact do talk like their friends. Alternatively, we could substitute continue to for still, giving a temporal aspect. The latter is the chosen function as still appears immediately before the verb talk. If this was meant to be contrastive, the writer would have used the collocation but still, followed by a comma. In example (10), still forms part of the adverbial phrase still in this century. If the writer wished to use still as a contrastive marker, they would have followed still with a comma, to show it was not a whole adverbial phrase. Example (11) could be interpreted as meaning the use of dam continues to feel pretentious. However, when one considers the context, the idea of dam feeling pretentious over time is infelicitous. Instead, the writer is contrasting the feeling of pretentiousness with the unwillingness to use kvinna. These analyses are subjective and perhaps give meaning to text that the author did not intend. However, when the context is ambiguous, other factors have to be considered, such as substitution tests, punctuation and word order, in order to derive meaning. It is hoped that these procedures accurately discerned the true function of still.

3.2.1.2 Determining yet as a Contrast Marker

Yet is an adverbial with a number of functions other than contrast marker. Yet can function as an adverb, meaning up to now, in the future, more and even now (Cambridge Advanced Learners Dictionary 2010). There were no examples in the subcorpora where there was ambiguity between a contrast marker and the other functions of yet. As a result, it was straightforward to mechanically divide the tokens of yet into contrastive and other.
3.2.2 Analyzing the texts at the ideational level for concession

It was not feasible to analyse all occurrences of however found in the subcorpora. There are 573 tokens of however in the KC subcorpus and 354 tokens of however in the SU LING subcorpus. Instead, a random ten percent sample of however tokens were chosen for analysis. The lower frequency of nevertheless/nonetheless, still and yet meant that larger samples were chosen, at random; ten examples of each for each subcorpus, apart from still in the KC corpus, which elicited only one example of usage. After the random sampling, each token was analyzed at the ideational level. The contrasted ideas were labelled S1 and S2 and analyzed to see if a concessive or non-

concessive relationship held between them. In naturally occurring discourse the difference between concessive and non-concessive relationships can be quite ambiguous. Therefore I used a systematic process to determine concession and nonconcession based on the Mann & Thompson model (1988). Using the Mann & Thompson model, a relationship is defined as concessive if a claim in S1 is conceded, but S2 can either be regarded as more important than S1, or S2 can cancel S1 to some degree. A relationship is defined as non-concessive if S1 and S2 are in balance, with a neutral stance taken, or if S1 and S2 are incompatible, with the writer taking a stance. In practice, this meant a search of the text to find the two ideas that were contrasted; in other words, finding S1 and S2. In example (12), S1 can be paraphrased as: When it comes to English teaching, attitudes are based on the views of people who act as if there is only one variety of English language. S2 is simply,there are many varieties of English. (12) Wilkinson (1995: 29) argues that attitudes to English teaching are based upon the views of people who are committed to an erroneously simplistic model of what the English language is. However, as we have discussed, there are many varieties of English. Therefore, Crowley (2003: 241) discusses the teaching of different varieties of English like we teach other languages such as French and German. Therefore no one variety is automatically superior. (KC 2 LING 28)

Then surrounding text was searched for evidence and clues on how the reader is meant to regard S1 or S2, or both. There is evidence in example (12) that the writer holds S1 to be false and therefore incompatible with S2; for example, the comment,[t]herefore no one variety is automatically superior and the deliberate inclusion of erroneously in the quote describing the single English model. In most cases, the writer gives clues to the nature of the contrastive relationship in the surrounding text. Of course, the surrounding text does not always give sufficient evidence to determine whether a concessive or non-concessive relationship holds. If so, the sense of the discourse marker was taken into consideration. Nevertheless/nonetheless, yet and still have a strong sense of concession (Fraser 1998; Bell 2010), so fuzzy examples with these discourse markers were analyzed as concessive. However has a concessive and non-concessive semantic sense (Ball 2010), which means that fuzzy examples cannot be treated in the same way as for nevertheless, still and yet. In fuzzy examples involving however, I was forced into looking more deeply into the text for an idea of concession or not. The last step of the process was using another linguist to corroborate or contradict the results for the largest groups. Although the preceding steps were systematic, there were some cases of disagreement. In cases of disagreement I returned to these examples and changed my analysis if the counter argument was convincing, or kept the same analysis if I found further evidence that supported my original analysis.

3.3 Co-occurrence

This study focussed on two forms of co-occurrence. First, paragraphs containing the contrast marker however were manually examined to find how frequently nevertheless/nonetheless co-occurred therein. Second, the collocation function in the AntConc software was used to find co-occurrence patterns of however (as this was the most frequently used contrast marker). AntConc calculates the frequency of words that appear within a given span of a token. A wide window span was used to find collocational patterns of however, from ten positions to the left and ten positions to the right. This wide search was used as the argumentative nature of academic essays can lead to the usage of extensive syntactic strings. Of course, limiting the window span to 10 spaces meant that examples of collocation were left out. However, an increased window span meant that unconnected ideas were included. Therefore, a window span of 10 places either side seemed to be an acceptable compromise. Example (13) shows how mainly is separated from however by seven words. (13) When children are about 1.11 years their speech consists mainly on one or two word utterances. However, some children are able to produce sentences consisting of about five words. (SU B LING 50) Mainly is used for a generalisation about childs speech, which is then signaled by however as not always being the case. The use of mainly is an example of hedging, a common feature of academic writing, where the author shows caution and modesty and, as a result, diplomatically develops the research space (Swales 1990). As hedging is a common feature of academic writing, commonly co-occurring hedges were specifically searched as well as other collocations, including other contrast markers.

4. Results
The results are presented in the following order. First there is a comparison of the use of however as a concession marker in the NS and NNS subcorpora. Second, the results of frequency of usage of nevertheless/nonetheless, still and yet as concession markers in both subcorpora are presented. Third, co-occurrence patterns of however and nevertheless/nonetheless in the same paragraph are described. Fourth, collocational patterns are presented. Last, I present observations of how concession is used within the academic essay by the NS and the NNS alike.
4.1 Use of however as a concessive and non-concessive marker

This study took a random 10% sample of however tokens to analyse whether a concessive or non-concessive relationship was signalled. The results from the samples investigated in this study (see table 2) show that both the NS and the NNS use however primarily for concessive purposes. Approximately 60% of the uses of however by

10

undergraduate students, NS and NNS, were signals of concession. In addition, there was one example in the other category in both subcorpora, as seen in examples (14) and (15); they were infelicitious in that they, respectively, signal the addition of further information and are used instead of whether. These other occurrences were so infrequent that infelicitious usage appears to be unworthy of further analysis. (14) As mentioned before nativism undermines the role of interaction. However there are other experiences that have been excluded from its elaborate explanations. (KC 2 LING 02)

(15) To investigate however these expressions are common and used by native speakers of English two methods was being used. (SU B LING 22)

The results show that when these particular NS and NNS use however, they use it in the same way. Therefore we can assume that the NS and the NNS have a similar understanding of the prototypical sense of however. We know that Swedish NNS will underuse however (e.g. Heino 2010), so it follows that they underuse however to signal concession.There are at least two possible explanations for this. First, Swedish NNS use fewer contrastive structures in their essays and therefore fewer contrast markers. It is beyond the scope of this study to confirm or deny this explanation and will be until such research is undertaken. Nevertheless, Swedish NS use contrast markers approximately as frequently as English NS, which suggests Swedish NNS have a problem with the contrast markers themselves when writing in English (Altenberg & Tapper 1998). This brings us to the second, preferred, explanation for Swedish NNS underusing however to signal concession. Although Swedish NNS have a similar grasp of the prototypical sense of however to NS, they do not have a similar grasp of its prototypical frequency of usage in signalling concession. However is the prototypical default option for signalling concession, and NS seem more aware of this than Swedish NNS. There are alternative contrast markers for signalling concession but their usage is prototypically infrequent in comparison to however. Usage of the alternatives is rationed, in the shadow of howevers prototypical centrality.

Table 2. Usage of however in random 10% samples of subcorpora KC Concession n % 38 66.7 Non concession 18 31.6 1 1.8 23 63.9 Other SU Concession Non concession 12 33.3 1 2.8 Other

11

4.2 Use of nevertheless/nonetheless, still and yet as concessive and non-concessive markers

The study took ten samples (apart from still in the KC subcorpus where only one example was found) at random from both subcorpora to analyze whether the contrast markers nevertheless/nonetheless, still and yet were used to signal a concessive or nonconcessive relationship. The results suggest that the NS use nevertheless/nonetheless solely for concessive purposes (see table 3). The NNS seem to be almost as aware as the NS that nevertheless/nonetheless has a prototypical sense of concession. There was one example of a NNS using nevertheless/nonetheless infelicitously, as seen in example (16). From this infelicitous use, we can see that Swedish NNS occasionally use nevertheless/nonetheless erroneously, but this is so seldom that it cannot be considered an important factor in explaining the NNS overuse of nevertheless/nonetheless. (16) It is clear that these differences occur everywhere but more so in New York City because everyone there has different histories and backgrounds, it is nevertheless a fairly new city. (SU B LING 58)

Table 3. Use of nevertheless/nonetheless, still and yet by subcorpus Marker Subcorpus Concession n Nevertheless Nonetheless Still KC SU LING KC SU LING Yet KC SU LING 10 9 1 9 6 8 % 100 90 100 90 60 80 PT 16 42 1 19 17 24 Non-concession n 0 0 0 1 4 2 % 0 0 0 10 40 20 PT 0 0 0 2 11 6 Other n 0 1 0 0 0 0 % 0 10 0 0 0 0 PT 0 5 0 0 0 0

* PT = Projected Total

The results also show that the NS and the NNS seem to have a similar grasp of the prototypical sense of still and yet, although the NNS use yet slightly more frequently than NS in signalling concession (80% of the time for NNS against 60% of the time for NS). This difference may be explained by yet having additional functions to signalling concession. Examples (17) and (18) show how yet is used to juxtapose ideas, or used in a similar way to on the other hand. Nevertheless, these results reflect Bells (2010) findings, which place nevertheless/nonetheless on a cline of concession with nevertheless/nonetheless the strongest and yet the weakest. Interestingly, yet has the weakest sense of concession but is used more frequently than still.

12

(17) The low number of instances of Ms can probably be attributed to the fact that it is a more modern yet less common term compared to Miss. (KC 3 LING 58) (18) The commonly use of' however has then omitted comparable expressions like: On the contrary, on the one hand ... on the other hand, despite, in contrast or nevertheless; expressions which are very rarely used or not used at all by the writers. Yet, beginning a sentence with but is frequent and appears six times in the texts. (SU B LING 66)

Like the results for however showed, it seems that while the Swedish NNS have a similar grasp of the prototypical sense of nevertheless/nonetheless, still and yet to the NS, they have a weaker grasp of the prototypical frequency of usage. If we project the sample results onto the subcorpora as a whole, we see that the NNS use nevertheless, still and yet to a greater degree than the NS. It seems to be the case that the Swedish NNS are less aware than the NS that nevertheless, still and yet have a prototypically low frequency of usage. This relative lack of awareness of frequency of usage would be even more marked if we found that Swedish NNS were using contrastive structures less frequently than NS. Moreover, if we accept that NNS will be less aware of the prototypical frequency of usage of nevertheless/nonetheless, still and yet, it seems logical to accept that NNS might be less aware of the prototypical frequency usage of however.
4.3 However and Nevertheless in the same paragraph

As nevertheless/nonetheless and however are the most frequently used concessive markers in the subcorpora, they have been analyzed for co-occurring usage in the same paragraph, and the results for the subcorpora compared. Co-occurrence is recorded when a paragraph containing a token of nevertheless/nonetheless also contains a token for however, as seen in example (19); this is only a part of the whole paragraph. (19) It will be noticed, however, that there are no red-top tabloids among those represented in my data. This may be considered an unfortunate omission in view of my suggestion that my data should be representative; nonetheless, my search returned no articles published in red-top newspapers on those days. (KC 2 LING 33) The results show differences in co-occurrence of however and nevertheless/nonetheless in the NS and the NNS paragraphs (see table 4). In the KC subcorpus approximately 50% of the paragraphs containing nevertheless/nonetheless also contain however. This compares with 20% of the paragraphs in the SU LING subcorpus containing nevertheless/nonetheless also containing however.

Table 4. Co-occurrence of however and nevertheless by paragraph

13

KC n of paragraphs containing nevertheless n of paragraphs containing both however & nevertheless 7 15

SU LING 44 9

% of paragraphs containing both however & nevertheless

47%

20%

These results are not conclusive. On the one hand, they could merely reflect the higher frequency of usage of however by the NS. Nevertheless/nonetheless co-occurs to a greater extent in paragraphs written by NS, as tokens of however are widespread and found in many paragraphs. On the other hand, these results could reflect the greater awareness NS seem to have of the prototypical high frequency of usage of however. Half the tokens of nevertheless/nonetheless occur in paragraphs containing however. It could be argued that part of the function of nevertheless/nonetheless is to avoid repeating however. Therefore nevertheless/nonetheless could serve two semantic, even rhetorical, roles. The first is when the author wants to be explicit in signalling concession; as we have already observed, nevertheless/nonetheless has the strongest sense of concession (Bell 2010). Nevertheless/nonetheless also has the role of an alternative concession marker that gives writing so-called elegant variation when however is used in proximity. The figure of 20% co-occurrence of however and nevertheless/nonetheless in the same paragraph in SU LING is an indicator that the NNS have a weaker sense of prototypical frequency of usage of nevertheless/nonetheless, not only overusing it as a concession marker but also underusing it in its alternative marker role, as seen in example (19).
4.4 Collocations & Hedging

Having looked at co-occurence of the most frequently occurring contrast markers within paragraphs, the next stage was to observe lexical co-occurrence patterns. During the process of analyzing the samples of text from the subcorpora it soon became apparent that certain lexical units were repeated. For example, contrast markers, such as although, were used in proximity to however. There were also examples of hedging (e.g. suggest) and a category labelled others which includes now, tradition, reality and current. The items in the other category represent concepts that were repeatedly contrasted. A constant theme seems to be temporal concepts, as the examples now, tradition and current show. Moreover, different collocates were included in the total, such as suggesting and traditionally; these are referred to as etc in Table 5. The collocation tool of the AntConc concordance software was used to find whether the NS and the NNS had different co-occurrence patterns in their use of however, specifically with other contrastors, hedges and the other category. The search for collocational patterns was limited to however as it is the only substantially occurring contrast marker. The results in table 5 show that there are some noteworthy differences in collocational usage between the NS and the NNS. For NNS the most striking

14

collocational feature is the relative overuse of but in comparison to the NNS. This could be due to the Swedish NNS relative unfamiliarity with formal registers (Altenberg & Tapper 1998). The NS seem to have a group of collocational partners they use with however that the NNS use to a much lesser degree. The contrast marker although is an illustration, as seen in example (20). In addition, the NS use appear, may and suggest to a greater degree than the Swedish NNS. The Swedish NNS seem to rely more heavily on seem for hedging in collocation with however. (20) Kachru ( 1991 ) however is a little more sceptical, stating that although language policies and other conscious efforts play their parts in the institutionalisation of a language, the process in reality is more organic, being more a matter of recognition of social, cultural, historical, and linguistic realities. (KC 3 LING 53)

Table 5. Number and frequency of collocation with however in a window span of 10L and 10R

Category contrastors although but though whereas hedges appear etc mainly may might possible etc seem etc suggest etc other current etc now reality tradition etc

KC n 25 4 2 3 16 2 24 3 9 27 19 3 8 3 5

% 4,4 0,7 0,3 0,5 2,8 0,3 4,2 0,5 1,6 4,7 3,3 0,5 1,4 0,5 0,9

SU n 8 21 2 3 4 5 9 0 8 20 5 0 1 1 0

% 2,3 5,9 0,6 0,8 1,1 1,4 2,5 0,0 2,3 5,6 1,4 0,0 0,3 0,3 0,0

Regarding the other category, it is difficult to come to any conclusions as both the NS and the NNS show little evidence of strong collocational patterns. Nevertheless, it can be said that all the lexical items in the other category are more likely to be used by the NS than the NNS; there were 19 examples of collocations in the other category by NS compared with just 2 for NNS. Therefore, these results do suggest that there are certain conceptual items that are repeated in contrastive relationships. It would make an interesting topic of study to analyze what these conceptual groups are and how frequently they occur in academic essays.

15

The results in Table 5 include all the instances of however in both subcorpora. The same process was not repeated for instances of concessive and non-concessive use of however, so it is not apparent whether there are any patterns regarding co-occurrence with however in its concessive and non-concessive function. Nevertheless, further study of this would be of considerable interest.
4.5 Concession within the academic essay

This research started with counting the frequency of usage of the studied contrast markers and then moved to a qualitative analysis of their use at the ideational level. The next stage was to observe patterns of co-occurence within the paragraph and finished with an analysis of the context of use, as described here. Two recurring contexts of use were apparent in this study; the use of concessive relationships to indicate a gap (Swales 1990: 141) and to delimit the extent of the essay as shown in examples (21) and (22) respectively. Examples of both features were found in the NS and NNS essays. While it is beyond the scope of this study to research whether NS use concession markers more than NNS in indicating a gap or delimiting the extent of the paper, it would make a fruitful area of future research. (21) The tip of the tongue phenomenon is one which most people have experienced at some point in their lives. Its can be defined as [QUOTE] Harley 2001. However there are no set explanations of the causes and effect of tip of tip of the tongue ... (KC 3 LING 08) (22) First of all we have to distinguish this aspect of language switch from the close related code-switching. Trudgill (2000, p. 106) describes codeswitching as a rapid process of switching from one language to another when the situation demands. In the Bilingualism Reader (2000, p. 16) Wei emphasizes that this occurs consciously to some extent and varies depending on the type of person addressed. This essay, however, focuses on the circumstances existing when bilingual people living abroad subconsciously react in their native language and, above all, when they are alone. (SU A LING 05)

5. Discussion
The results show that the NS use however to a greater degree than the NNS to signal concession. On the other hand, the NNS use nevertheless/nonetheless, still and yet to a greater degree than the NS in signalling concession. In one respect, these results could be considered contradictory. Less frequent use of contrast markers is a reflection of proficient and mature writers of English (Crewe 1990; Shaw 2009; Tapper 2005). The NNSs overuse of nevertheless/nonetheless, still and yet is possibly evidence of lesser profiency in writing in English. On the other hand, I will argue that the NNS underuse

16

of however in signalling concession is a sign of lesser proficiency in writing in English, as this underuse is likely to be a result of stylistic uncertainty and native language transfer.
5.1 Stylistic Uncertainty

Differences in usage of contrast markers by NS and Swedish NNS can be attributed to lack of familiarity with formality (Altenberg & Tapper 1998). In evidence Altenberg & Tapper cite the overuse of but by Swedish NNS, and underuse of the more formal however by Swedish NNS. This study also provides results that seem to agree with the lack of formality hypothesis. The Swedish NNS are more likely to use the informal but in collocation with however, whereas the NS are more likely to use the more formal although (see table 5). The Swedish NNS overuse of the formal nevertheless/nonetheless (see table 3) makes Altenberg & Tappers (1998) idea of general stylistic uncertainty more descriptive than their idea of lack of familiarity with formality hypothesis. Suitable writing style entails familiarity with formality, but it also entails familiarity with prototypical frequency of usage. Although the Swedish NNS seem to have a similar idea of the prototypical sense of the contrast markers analyzed in this study to NS, these results seem to show the Swedish NNS have a weaker grasp of the prototypical frequency of usage. Swedish NNS signal concession differently to NS, underusing however and overusing nevertheless/ nonetheless, still and yet. NS appear to have a subconscious understanding of the prototypical centrality of however. NS also seem to have a deeper understanding that nevertheless/nonetheless (as well as yet and still) is far removed from prototypical centrality. It has a prototypical sense as the strongest concession marker (Bell 2010), which perhaps entails that it should be used rather sparingly. Therefore when nevertheless/nonetheless is used it has greater rhetorical effect. In this way we can see that proficient usage involves more than understanding the sense of the contrast marker. There are also concepts such as frequency of usage, which are harder for NNS to learn than a contrast markers prototypical sense. The learning process could explain the NNS stylistic uncertainty in using contrast markers. Typical textbook exercises, such as the substitution exercise in example (23), are of little use as they do not consider prototypical frequency of usage. Moreover, typical textbook exercises do not show how contrast markers should only be used to aid the logical development of arguments (Crewe 1990). (23) Rewrite each of the following sentences using the words given. Make any necessary changes to punctuation. 2. The wind blew all the time, but we still managed to enjoy ourselves. (nevertheless). (Side & Wellman 2002:107)

17

This example is not very helpful to L2 students as it treats contrast markers as lexical items that merely replace another item. For example, we have seen that however cooccurs fairly regularly with other contrastors such as although, or with hedges such as may (see Table 5), but exercises such as example (23) pay no attention to this. Instead of using contrast markers to decorate a text, they should only be used if strictly necessary, to help with cohesion or as a rhetorical device; L2 students should be made aware of the primary importance of logical development of arguments. In addition, teachers should be aware of the noticing technique (Thornbury 1997). In the noticing technique, teachers help students notice and deduct how contrast markers operate by highlighting how they are used in naturally occurring texts.
5.2 Native Language transfer

Previous studies have found that Swedish NNS overuse still (Altenberg & Tapper 1998, Heino 2010, Tapper 2005), which was again found to be the case in this study (see Table 3). The overuse of still by can be described as resulting from native language transfer, which is universally accepted to play a role in foreign language production (Granger & Tyson 1996). Evidence of the role of native language transfer can be found in the Swedish English bidirectional translation corpus by Altenberg (2002). Altenbergs study compared the use of English and Swedish contrast markers. A bi-directional translation corpus was used, showing empirically how the same contrastive meaning is expressed in Swedish and English. Altenbergs study found a number of distinct subsets of contrastive relationships between Swedish and English contrast markers, two of which are relevant for this study. One of the subsets was labelled concessives, and it includes nevertheless, yet and still and the Swedish equivalent nd. The other relevant subset was labelled adversatives, and it includes however and the Swedish equivalents emellertid or dock. Altenberg used Frasers (1998: 313-315) descriptions of the senses of however and on the other hand to define adversative. The Altenberg study showed that Swedish NS primarily use nd to signal concession. In the bi-directional translation study nd had a relatively strong mutual correspondence to still, nevertheless or yet. Although Altenbergs study concerned professional translations, its results are relevant to this study of university students. Altenberg showed that nd has a strong sense of concession in Swedish and that nd has a strong mutual correspondence to still. Therefore, it is possible that some Swedish NNS in signalling concession are transferring the sense of nd and its strong mutual correspondence to still and hence are overusing still. If we accept that one difference in usage (still) is due to native language transfer, it follows that other differences in usage, the overuse of nevertheless/nonetheless and yet by Swedish NNS, could also at least partly be due to transfer from Swedish to English. The Altenberg study showed there was no mutual correspondence, not even weak correspondence, recorded between nd and however. However had mutual correspondence with a non-concessive group of contrast markers: emellertid, dock and dremot. As Altenberg showed, nd is the contrast marker with the strongest sense of concession, yet it has no mutual correspondence with however. This aspect of native

18

transfer could explain why Swedish NNS have a lesser grasp of the prototypical centrality of however to signal concession. Interestingly, nowhere does Altenberg express surpise at professional translators ignoring howevers prototypical sense of concession, but this could be due to a different definition of concession, as described in section 2.2. The evidence from the bi-directional translation study shows that the most frequently used concession marker nd is more commomoly translated into nevertheless/ nonetheless, still and yet than however. It is likely that often Swedish NNS, even advanced writers, are following a line from concession to nd to nevertheless/ nonetheless, still and yet. It appears the lack of awareness of the prototypical centrality of however is also likely to result from this native language transfer.

6. Conclusions
The results of this study suggest that NS use however to a greater degree than Swedish NNS for signalling concession. Instead of using however as frequently as NS, Swedish NNS use nevertheless/nonetheless, still and yet more frequently than NS. These findings only partially answer the question of what Swedish NNS do instead of NS when signalling concession. Indeed, this question forms the backbone of suggested further research. I would suggest that further research takes the form of analyzing samples of complete essays from both subcorpora. The essays could be analyzed to confirm that the NS and the Swedish NNS use contrastive structures to the same extent. Moreover, the logical organisation of contrasting arguments could be studied and compared between subcorpora. The essays could be analyzed to see how the students signal contrastive relationships, especially concessive relationships (including omitted signalling altogether). This study has examined a handful of contrast markers, but there are many more markers that signal contrastive relationships, such as although, though, despite, while, whereas, and notwithstanding or other structures such as the fixed phrase be that as it may. This survey has also attempted to discover the reasons for underuse of however and overuse of nevertheless/nonetheless, still and yet by Swedish NNS. It seems that Swedish NNS have a weaker grasp of the prototypical centrality of using however to signal concessive relationships. In contrast, Swedish NSoveruse of nevertheless/ nonetheless might be explained by their lack of awareness of nevertheless/nonethelesss prototypical role. It appears that nevertheless/nonetheless is often used for variety, or as a more rhetorical device to signal concession. I have suggested possible causes for Swedish NNS weaker sense of contrast markers prototypical frequency of usage. It seems that native language transfer from Swedish to English is likely to have an effect. At the same time, teaching materials and linguistic resources should be re-evaluated. It is argued that a change is needed in these teaching materials and linguistic resources, as suggested by Crewe. Instead of translating contrast markers, or doing substitution exercises in text books, it would be far more helpful for

19

L2 students to study the logical progression of arguments. Although these recommendations result from comparing Swedish NNS with NS, I tentatively suggest that these recommendations would be useful for all L2 students. The recommendations above would hopefully be beneficial in helping writers to structure their arguments and in helping readers to follow the essays logic. After all, a reader of an essay is like a motorist in traffic. The motorist feels comfortable if the cars on the road are signalling clearly. It is confusing if a car signals when it is not necessary. It can be disconcerting if a car makes a turn without signalling. It is the same for the reader of an essay. Arguments should be clear and easily followed. The reader should not be confused by unnecessary use of contrast markers or sudden umarked changes in an arguments direction. Correct but restrained use of contrast markers can only help the reader of the essay.

20

References
Altenberg, B. (2002). Concessive markers in English and Swedish. In H. Hasselgrd, S. Johansson, B. Behrens & C. Fabricius-Hansen (Eds.), Information Structure in a Cross- Linguistic Perspective, 21-43. Amsterdam: Rodopi. Altenberg, B. & Tapper, M. (1998). The use of adverbial markers in advanced Swedish learners written English. In S Granger (Ed.), Learner English On Computer. London and New York: Longman. Anthony, L. (2007). AntConc (Version 3.2.1w Windows. Retrieved October 13, 2010, from http://www.antlab.sci.waseda.ac.jp/software.html Bell, M (2010). Nevertheless, still and yet: Concessive cancellative discourse markers. Journal of Pragmatics 42 (7), 19121927. Cambridge Advanced Learner's Dictionary. (2010). Cambridge University Press. Retrieved December 16, 2010, from http://dictionary.cambridge.org/search/british/?q=yet Crewe, W.J. (1990). The illogic of logical markers. ELT Journal, 44(4), 316325. Fraser, B. (1998). Contrastive discourse markers in English. In A.H. Jucker & Y. Ziv (Eds.) Discourse Markers. Descriptions and Theories, 301-326. Amsterdam: John Benjamins. Granger, S. & Tyson, S. (1996). Connector usage in the English essay writing of native and non-native EFL speakers of English. World Englishes 15, 1727. Heino, P (2010). Adverbial Markers in Advanced EFL Learners and Native Speakers Student Writing. Unpublished BA Essay, English Department, Stockholm University, Spring 2010. Mann, W & Thompson, S.(1988). Rhetorical Structure Theory: Towards a Functional Theory of Text Organization, Text 8 (3), 243-281. Quirk, R., Greenbaum, G., Leech, G. & Svartvik J. (1985). A Comprehensive Grammar of the English Language. London: Longman. Side, R & Wellman G. (2002). Grammar and Vocabulary for Cambridge Advanced and Proficiency. Harlow: Longman. Shaw P, (2009). Linking Adverbials in Student and Professional Writing in Literary Studies: What Makes Writing Mature. In Charles, M. et al Academic Writing: At the Interface of Corpus and Discourse (pp. 215235). London: Continuum. Swales, J.M. (1990). Genre Analysis: English in Academic and Research Settings. Cambridge: Cambridge University Press. Tapper, M. (2005). Connectives in advanced EFL Learners written English preliminary results. In F. Heinat & E. Klingvall (Eds.), The Department of English in Lund: Working Papers in Linguistics 5. Lund: Department of English, Lund University. Thornbury, S (1997). Reformulation and Reconstruction: Tasks that Promote 'Noticing'. ELT Journal, 51(4), 326-335.

21

Stockholms universitet 106 91 Stockholm Telefon: 0816 20 00 www.su.se