Sentence Connectors in English Academic Writing

Philosophische Fakultät
Professur Englische Sprachwissenschaft
Magisterarbeit
Sentence Connectors in English Academic Writing An Empirical Comparison of Research Articles by German
and Native English Writers
eingereicht von
Ellen Wießner
geboren am 14. August 1982
Betreuer: Prof. Dr. Josef Schmied
Dr. Christoph Haase
Chemnitz, den 7. Juli 2008
ii
Contents
List of Abbreviations
iv
List of Figures
v
List of Tables
vi
1 Introduction
1
2 Academic Writing and the EFL Context
4
2.1
Academic English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
2.2
Writing Across Cultures . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.3
English as a Foreign Language . . . . . . . . . . . . . . . . . . . . . . . . .
13
2.3.1
Aspects on L2 Composition and Assessment . . . . . . . . . . . . . .
15
2.3.2
Previous Research on Cohesion in EFL Texts . . . . . . . . . . . . .
18
3 Sentence Connectors - Functions and Uses
22
3.1
Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22
3.2
Structural Relations between Clauses . . . . . . . . . . . . . . . . . . . . . .
23
3.2.1
Coordination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
3.2.2
Subordination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
3.2.3
Adverbial Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
3.2.4
Overlap of Syntactic Categories . . . . . . . . . . . . . . . . . . . . .
27
Logical Relations between Clauses . . . . . . . . . . . . . . . . . . . . . . .
29
3.3.1
Additive Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
3.3.2
Adversative Relation . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
3.3.3
Causal Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
3.3.4
Temporal Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
3.3.5
Overlap of Semantic Categories . . . . . . . . . . . . . . . . . . . . .
32
Position of Sentence Connectors . . . . . . . . . . . . . . . . . . . . . . . . .
34
3.3
3.4
Contents
iii
3.5
35
Sentence Connectors and Academic Writing . . . . . . . . . . . . . . . . . .
4 Empirical Study
38
4.1
Purpose of the Study and Research Questions . . . . . . . . . . . . . . . . .
38
4.2
Corpus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
4.3
Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
4.4
Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
4.5
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
4.5.1
Overall Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
4.5.2
Overall Variety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
46
4.5.3
Syntactic Categories . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
4.5.4
Semantic Categories . . . . . . . . . . . . . . . . . . . . . . . . . . .
50
4.5.5
Connector Position . . . . . . . . . . . . . . . . . . . . . . . . . . . .
58
5 Summary of Findings and Discussion
60
5.1
Main Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
5.2
Connector Frequency: A Pointer to L1 Interference? . . . . . . . . . . . . .
61
5.3
Connector Variety: Native and Non-Native Preferences . . . . . . . . . . . .
61
5.3.1
’Lexical Teddy Bears’ . . . . . . . . . . . . . . . . . . . . . . . . . .
62
5.3.2
Inconsistency in Register . . . . . . . . . . . . . . . . . . . . . . . .
62
5.3.3
British and American Preferences . . . . . . . . . . . . . . . . . . . .
63
Connector Position: A Pointer to Non-Native Pattern? . . . . . . . . . . . .
64
5.4
6 Conclusion and Implications for Future Research
67
Primary Sources
viii
Secondary Sources
xi
A Additional Data on Primary Material
xv
B Additional Tables
xxii
Zusammenfassung
xxviii
Selbstständigkeitserklärung
xxxiv
iv
List of Abbreviations
Abbreviation
Meaning
AE
American English
BE
British English
CConj
Coordinating Conjunction
EAP
ESL/EFL
English for Academic Purposes
English as a Second/Foreign Language
L1
L2
LinkAdv
First or Native Language
Second or Foreign Language
Linking Adverbial
MELAB
Michigan English Language Assessment Battery
RA
Research Article
SConj
Subordinating Conjunction
TOEFL
Test of English as a Foreign Language
v
List of Figures
4.1
Distribution across Syntactic Categories . . . . . . . . . . . . . . . . . . . .
47
4.2
Distribution across Semantic Categories . . . . . . . . . . . . . . . . . . . .
50
4.3
Sentence-Initial Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
vi
List of Tables
4.1
Statistical Data of Text Corpus . . . . . . . . . . . . . . . . . . . . . . . . .
41
4.2
Overall Connector Frequency . . . . . . . . . . . . . . . . . . . . . . . . . .
45
4.3
Most Frequently Used Connectors . . . . . . . . . . . . . . . . . . . . . . .
46
4.4
Connector Variety according to Syntactic Categories . . . . . . . . . . . . .
48
4.5
Clause-level Coordinators . . . . . . . . . . . . . . . . . . . . . . . . . . . .
48
4.6
Coordinators and Linking Adverbials in Combination . . . . . . . . . . . . .
49
4.7
Additive Linking Adverbials . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
4.8
Adversative Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
4.9
Selected Adversative Subordinators . . . . . . . . . . . . . . . . . . . . . . .
53
4.10 Causal Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
4.11 Temporal Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
57
4.12 Connectors with Multiple Semantic Roles . . . . . . . . . . . . . . . . . . .
57
4.13 Initial Connector Usage across Categories . . . . . . . . . . . . . . . . . . .
58
1
1 Introduction
Although much attention, time, effort, and resources are devoted to teaching L2 academic writing and its conventions, it appears that non-native speaker (NNS) text differs
substantially from that of native speakers (NSs) of similar academic standing (Hinkel,
2002, p. xvii).
As English is the world’s predominant language of research and scholarship (Swales, 1990),
scholars and scientists from all over the world are required to have proficient knowledge
of the conventions of English academic discourse these days. Especially when publishing
academic papers and literature, well-developed writing skills are necessary to ensure that
the intended meaning is communicated properly. Non-native speakers of English, however,
are challenged by writing in a foreign language. Studies have shown that there are substantial linguistic and rhetorical differences between native and non-native written texts
and that non-native speakers of English often lack coherence in their writing (see, for example, Hinkel, 2002; Silva, 1993; Wikborg, 1990). As has been emphasized by numerous
researchers, this lack of coherence “can seriously affect the credibility of non-native writers”
(Mauranen, 1993, p. 2) and thus may be detrimental to their academic and professional
careers.
The fact that non-native English writing is often conceived as incoherent and ineffective
by native speakers is partly attributed to the different writing styles and organizational
patterns employed by native and non-native English writers (Hinds, 1990; Mauranen, 1993),
but also to a deficient and inadequate use of cohesive ties (e.g., Hinkel, 2002; Milton and
Tsang, 1993). The connection between cohesion and coherence is in fact an often discussed
issue in textlinguistic research. It is evident, however, that both cohesion and coherence are
crucial to produce clear and well-organized texts. Whereas cohesion is generally referred
to the surface structure of a text, including overt semantic relations linking sentences or
sections of a text together (Halliday and Hasan, 1976), coherence mostly derives from
the concepts and relations underlying the sentences (DeBeaugrande and Dressler, 1981).
Cohesion thus can be defined as “continuity in word and sentence structure” and coherence
1. Introduction
2
as “continuity in meaning and context” (Louwerse and Graesser, 2005, p. 216).
Although the mere presence of cohesive devices is no guarantee for a text to be coherent,
it has been shown that cohesion contributes to coherence in various ways and that the
absence or misuse of explicit cohesive markers may rather distract the reader and lead to
what Bublitz and Lenk (1999) refer to as ’disturbed coherence’. Moreover, as Lorenz (1999)
points out, “advanced learners’ deficits are most resilient in the area of lexico-grammar,
where lexical items are employed to signal grammatical and textual relations” (p. 56). This
clearly involves cohesive ties, such as conjunctive expressions and demonstrative reference,
and should therefore be subject to further consideration.
Most textlinguistic studies analyzing cohesive ties in foreign language writing are based
on the concept of cohesion formulated by Halliday and Hasan (1976). In their work on
text-internal cohesion in English, they distinguish five types of cohesive relations, namely
reference, substitution, ellipsis, conjunction and lexical cohesion. Probably the most distinctive feature of non-native writing is that of conjunction. Conjunctive expressions include linking words and phrases from different grammatical categories, i.e. coordinators,
subordinators and linking adverbials, and are generally subsumed under the umbrella terms
connectors or connectives. Many researchers stress the importance of connectors for marking logical relations between propositions (cf. Halliday and Hasan, 1976; Greenbaum and
Quirk, 1990; Leech and Svartvik, 1994) and, according to Lorenz (1999),
connector usage is evidently an area which is sensitive to markedly non-native style.
Whichever way a given group of learners may lean in terms of connector frequency, and
whether or not the number of connectives correlates with proficiency, the way in which
logical relations are signaled does have significant bearing on the subjective coherence
within learner’s writing (p. 57).
One way to investigate connector usage in native and non-native writing is the comparative analysis of text corpora. Corpus-based studies are a useful tool in second language
research since they provide empirical evidence about actual language use. The quantitative and functional analysis of specific linguistic items in naturally-occurring texts allows to
identify major patterns in language which can be particularly helpful for language learning
and teaching. Thus “knowing when to use structures appropriately is an essential part of
developing communicative competence in a language” (Biber et al., 1998, p. 80), and so
the findings of corpus investigations can be used, for example, to inform writing instruction
and pedagogy.
1. Introduction
3
Research on non-native English writing, however, has been carried out predominantly
for speakers of English as a second language (ESL) living or studying in English-speaking
countries, i.e. the USA, Canada, Australia and the UK (Grabe and Kaplan, 1996). As
the majority of ESL students in these countries originate from non-Western cultures, studies on the English writing of speakers of Chinese (e.g., Field and Yip, 1992; Milton and
Tsang, 1993; Hinkel, 2002), Korean and Japanese (e.g., Hinkel, 2002) or Arabic (e.g., Reid,
1992; Hinkel, 2002) have taken a prominent place in contrastive and textlinguistic research.
In the European countries, research on English as a foreign language (EFL) composition
most notably includes EFL student writing from the Nordic countries, i.e. Finland, Sweden, Denmark and Norway, as reported in the NORDWRITE project (see, for example,
Evensen, 1990; Wikborg, 1990) as well as scientific writing by Finnish academics (Ventola
and Mauranen, 1991). The English writing of speakers of other European languages, such
as German, French and Italian as well as the East-European languages, has only been
addressed to a limited extent so far.
The purpose of this paper thus is to compare the English writing of German and native
English scientists. In particular, it is to furnish information about the frequency and
diversity of sentence connectors employed in English research articles by German and
native speakers of English. The study draws on the assumption that German EFL writers’
use of sentence connectors is different from that of native English speakers and that they
may over- or underuse these cohesive devices to a certain extent. It is further assumed
that native and non-native English writing differs with regard to connector position and
variety, and that these differences account for markedly native and non-native patterns of
connector usage.
The paper is organized in three main parts. The first part discusses some key issues
in English academic writing and provides an overview of previous research on writing in
a foreign language. The following chapter gives a syntactic and semantic classification
of sentence connectors providing a functional basis for the subsequent study. The third
part then presents the empirical comparison of sentence connectors in the English research
articles of German and native English writers, followed by a discussion of the main findings
and a final conclusion.
4
2 Academic Writing and the EFL Context
2.1 Academic English
Academic success usually depends on following the conventions for academic English
in the relevant field of study (Hoadley-Maidment and Mercer, 1998, p. 291).
English is used as a medium for communication in higher education not only in Englishspeaking, but also increasingly in non-English-speaking countries, and it serves as the lingua
franca in international scientific discourse (Hyland, 2006). Academic English therefore is of
crucial importance to the academic world. Though ’academic’ is a rather vague notion, in
text and discourse analysis, it is clearly associated with a particular type of discourse. “As
certain discourses become more deeply embedded in the social functioning of groups, these
discourses become conventionalized; they become recognized genres which serve functional
purposes in communication” (Grabe and Kaplan, 1996, p. 136).
Genres are characterized in terms of rhetorical features that are considered to be appropriate for a particular discourse. Disregarding those rhetorical conventions can have
serious effects on the speakers’ or writers’ reliability. If writers fail to meet the readers’
expectations of a particular text type, their statements are likely to be misconceived and
dismissed (Hinds, 1990; Mauranen, 1993). That is why EAP (English for Academic Purposes) programs aim at raising the speakers’ and writers’ awareness for the syntactic and
lexical particularities of academic and scientific discourse as well as for the different genres
and registers1 existing in academic English.
1
The terms ’genre’ and ’register’ are generally both used to describe functional language varieties serving
particular communicative purposes. Some linguists draw a distinction between the two in that ’genre’
is used to refer to the text type itself (e.g. a scientific research article) while ’register’ refers to the
stylistic choices (e.g. the language of science) (Swales, 1990). Instead of distinguishing between the
two concepts, most researchers, however, show a clear preference for either of the two, disregarding the
other. Swales (1990), for example, exclusively uses ’genre’ while Biber et al. (1999) exclusively uses
’register’. In the present study, both terms are used without particular specification.
2. Academic Writing and the EFL Context
5
Schmied (2007) also distinguishes between ’student EAP’ and ’research EAP’. Referring
to texts produced by students for their teachers, student EAP generally incorporates the
(re)production of information that is already known to the reader, i.e. the teacher, in the
form of student essays, term papers or oral presentations. Research EAP, in contrast, refers
to the exchange of specialized information about novel findings, methods and developments
between specialists of a particular scientific field. Genres of ’Research English’ include
doctoral dissertations, laboratory reports, case studies and research articles. As the study
in hand is concerned with the latter, that is, English scientific writing by academic experts
for academic experts, in the following, the genre of the research article is to be considered
in more detail.
The Genre of the Research Article
Scientific papers are a common text type in academic discourse. They distinguish themselves from other genres and registers by a number of linguistic properties. Biber (1995,
1988), for example, points to the very informational and abstract character of academic
prose in contrast to, e.g., newspaper language or conversation. He further indicates that
science and research articles (also ’research papers’ or ’research reports’) are highly specialized registers, that is, they are produced for very specific purposes and an increasingly
specialized readership (Biber, 1995, p. 300). According to Swales (1990), the research
article (RA) is a written text
usually limited to a few thousand words, that reports on some investigation carried
out by its author or authors. In addition, the RA will usually relate findings within it
to those of others, and may also examine issues of theory and/or methodology. It is to
appear or has appeared in a research journal or, less typically, in an edited book-length
collection of papers (p.93).
Hence, research articles are generally carefully planned and have been edited and revised
before they were published, which may entail certain adjustments to the editors’ stylistic
preferences.
Regarding the historical development of research papers, Swales (1990) observes “a shift
from description to explanation, thus suggesting increasing intellectual complexity” (p.115).
In fact, the primary purpose of scientific papers is the detailed explication and argumentation of research activities. Therefore, they generally feature a high degree of lexical
and syntactic complexity. In contrast to spoken registers, information is condensed by an
2. Academic Writing and the EFL Context
6
intensive use of nominalizations, adjectives and complex verbs, as well as complex noun
and prepositional phrases (Halliday, 2004; Biber, 1995, 1988). This dense integration of
information, however, often results in abstract and ambiguous text structures interfering
with the clarity and explicitness appreciated in scientific discourse. Halliday (2004), for
instance, identifies seven difficulties that are characteristic of scientific English:
1. Interlocking definitions,
2. Technical taxonomies,
3. Special expressions,
4. Lexical density,
5. Syntactic ambiguity,
6. Grammatical metaphor,
7. Semantic discontinuity (p.162).
In addition to this, the intelligibility of scientific texts might be compromised by the fact
that written communication is not interactive and does not allow for the clarification of
ambiguities or misconceptions by a direct reader-writer interaction as in spoken discourse.
Instead, the writer has to draw on rhetorical devices of written communication to convey
his or her ideas and to express attitudes and opinions in a clear and unambiguous manner.
“The greater complexity [of written registers] is generally attributed to two distinctive
characteristics of writing: the lack of strict time constraints during production, and the
need to establish cohesion strictly through the lexical-syntactic channel” (Biber, 1988,
p. 48). This is one reason why sentence connectors play an important role in the production
of scientific papers (see section 3.5).
With regard to macrostructure, Swales (1990) refers to the four standard sections of
research articles - Introduction, Method, Results and Discussion - suggesting that “the
different sections perform different rhetorical functions and thus require different linguistic
resources to realize those functions” (p. 136). Investigating the distribution of linguistic
and rhetorical features across these sections, he found that the Method and Results sections
incorporated fewer and simpler linguistic features, while the Introduction and Discussion
sections were much more complex in terms of functional and linguistic properties.
Similarly, the characteristic features of scientific discourse are not necessarily uniform
across disciplines:
There are systematic linguistic differences associated with the distinction between
the ’professional’ disciplines (business and engineering) and the more traditionally
2. Academic Writing and the EFL Context
7
’academic’ disciplines (humanities and social sciences). The ’professional’ disciplines
emphasize the mastery of technical methods and procedures, while the ’academic’ disciplines emphasize the discovery of new knowledge, the critical evaluation of information,
and discussion of alternative perspectives. These differences have strong linguistic correlates, for both vocabulary and grammatical features (Biber, 2006, p. 226).
As a consequence, science and engineering articles are found to be much more impersonal than papers written in the humanities and social sciences (Hyland, 2006), and Biber
(1995) found features of an abstract writing style, such as a very frequent use of conjuncts,
adverbial subordinators and passive constructions, “especially prominent in the academic
subregister of technical and engineering prose” (p.164). In addition, Swales and Feak
(2004) report that, in the social sciences, method sections are often described in considerable detail (the development of a new method sometimes even being the main point of the
RA), whereas engineering and the natural sciences more often draw on standard practices
and methods which are largely taken for granted. This leads to great variation in method
sections ranging from condensed methodologies (featuring, e.g., few linking phrases) at one
end of the spectrum to extended methodologies (featuring, e.g., a wide range of linking
phrases) at the other (p. 227). The organization and rhetorical style of RAs may be likewise diversified even within one scientific discipline depending on the writers’ individual
intentions and stylistic preferences.
2.2 Writing Across Cultures
Dealing with the specific register of academic writing, which is of particular interest to
university students and scholars around the world, there is yet another field of research to
receive special attention. Contrastive rhetoric, in addition to genre analysis investigating
language variation across genres and registers, is assuming that writing practices are also
culturally determined and that writers from different language backgrounds follow different
discourse patterns in their texts. Comparing the writing of individuals from various cultural
backgrounds, contrastive rhetoric has been particularly useful in second language (L2)
and EAP research as studies have shown that culture-specific writing patterns can cause
interference in L2 writing (Connor, 1996; Hinds, 1990; Clyne, 1987).
Studies contrasting linguistic and rhetorical features of academic genres across cultures
indicate that academic English tends to
• Be more explicit about its structure and purposes,
2. Academic Writing and the EFL Context
8
• Employ more, and more recent, citations,
• Use fewer rhetorical questions,
• Be generally less tolerant of digressions,
• Be more tentative and cautious in making claims,
• Have stricter conventions for sub-sections and their titles, and
• Use more sentence connectors (such as therefore and however)
as compared with other languages (Hyland, 2006, p. 44). Even within the (native) English
language, the influence of culture on writing practices is apparent. In a study comparing
British and American written genres, Biber (1987) revealed systematic differences between
the two English dialects: The British written genres were consistently more edited and
less interactive than the corresponding American genres, while the American genres were
consistently more colloquial and involved, but also more abstract than the British ones.
“The study suggests that the differences [...] are associated with greater attention to
grammatical and stylistic prescriptions in British than American writing” (Biber, 1987,
p. 99).
As stated earlier, academic success is also dependent on how far the rhetorical conventions of the target language are adhered to. Students and scientists writing in English as
a foreign language have to be aware of the different discourse strategies employed in the
respective English variety and their own native language before they “can begin to understand what [they] must do in order to write in a more native-like manner (or in a manner
that is more acceptable to native speakers of the target language)” (Grabe and Kaplan,
1996, p. 198). As a consequence, the body of contrastive studies uncovering rhetorical
differences between particular languages has grown substantially in the past decades.
Clyne (1987), for instance, has described organizational differences between English and
German academic discourse. Comparing linguistic and sociological texts written by English
and German speakers, he found that German texts tend to be characterized by content
digressions and discontinuity leading to a less linear organization than provided in English
texts. The study further revealed that English writers are more likely to employ advance
organizers explaining the organization of the paper, while the German texts featured a
much higher lexical density than the English ones. From that he concluded that
texts by Germans are less designed to be easy to read. Their emphasis is on providing readers with knowledge, theory, and stimulus to thought. [...] In both cultural
traditions, there is a co-operation between author and reader, with the author engaged
in [...] an elaborative process and the reader’s perspective being a reductive process. In
2. Academic Writing and the EFL Context
9
English-speaking countries, most of the onus falls on writers to make their texts readable, whereas it is the readers who have to make the extra effort in German-speaking
countries so that they can understand the texts, especially if the author is an academic
(Clyne, 1987, p. 238).
In addition to the German texts, Clyne’s analysis also incorporated English texts written by native speakers of German. Concerning the writing in EFL, the German writers
reported to be “aware of the difficulties they are encountering with English discourse patterns” (p.233). Nevertheless, supporting theories of native language (L1) interference in
L2 writing, Clyne (1987) found that “English texts by German scholars tend to contain
the same cultural discourse patterns as German texts” (ibid).
Furthermore, there are numerous studies2 indicating that oriental languages, such as
Chinese, Japanese and Korean, follow a rather indirect and implicit writing style leaving
reasoning and interpretation up to the reader (just as Clyne has suggested for academic
German). Due to the marginal presence of explicitness and divergent strategies to achieve
coherence, oriental writing seems to differ greatly from English discourse patterns.
In fact, analyzing expository writing in Japanese, Korean, Chinese and Thai, Hinds
(1990) found that these Asian languages all prefer a ’delayed introduction of purpose’ and
the organization of information from specific to general. From a Western perspective, this
writing style would be interpreted as inductive (thesis statement in the final position).
Although implied, the thesis statement in oriental writing is often not stated explicitly.
Hinds therefore suggests to term this organizational pattern ’quasi-inductive’. Native English speakers, however, are neither familiar with such an oriental quasi-inductive style nor
do they expect an essay to be organized inductively as defined in the Western tradition.
Rather, they expect the thesis statement in the initial position and the organization of text
from general to specific, i.e. a deductive style. As a result, “this delayed introduction of
purpose has the undesirable effect of making the essay appear incoherent to the Englishspeaking reader, although this style does not have this effect on the native speaker” (Hinds,
1990, p. 98).
Contrastive analyzes were also conducted for other languages, such as Finnish, Spanish,
Arabic and Hindi (cf., for example, Grabe and Kaplan, 1996; Connor, 1996). In her paper
“Reference in Academic Rhetoric: A Contrastive Study of Finnish and English Writing”
(1992), Mauranen found that Finnish scientists prefer a relatively late introduction of
2
A comprehensive overview of studies contrasting English with oriental and other languages is given by
Connor (1996).
2. Academic Writing and the EFL Context
10
central points and an inductive organizational style. Moreover, cohesive devices such as
demonstrative references were employed less frequently in the Finnish texts than in English
writing (Connor, 1996, p. 50f). In addition, some other research suggests that Finnish
writing differs greatly from English discourse patterns while it is closer to German writing
practices (ibid).
This raises another question contrastive rhetoric is concerned with: Are languages which
are closer in culture and/or share one language family (e.g. the Indo-European languages
English, German and Spanish) closer in their writing styles and organizational patterns
than languages that are culturally and linguistically very distinct from each other (e.g.
English vs. Arabic or Chinese)? The fact that many Asian languages were found to be
indirect and reader-responsible, quite contrary to English writing being said to be explicit
and writer-responsible, suggests that there is a correlation between cultural, linguistic
and rhetorical distance between languages. However, the findings of Clyne (1987) and
Mauranen (1992, in:
Connor, 1996, p. 50f) clearly indicate that German and Finnish
texts are rather reader-responsible and not as explicit as English texts, although both
these languages belong to the Western culture and German even is a Germanic language
just as English. Regarding the use of cohesive devices such as coordinating conjunctions,
Reid (1992) also found that the English essays of Spanish writers more closely resembled
the English writing of Arabic speakers than those by native speakers of English.
Findings like the above indicate that the role of the native language in L2 composition
cannot be clearly specified. What is clear, however, is that writing patterns are culturespecific and that L1 interference “interact[s] in complex ways with other factors” (Ellis,
1992, p. 33). Though contrastive rhetoric is frequently used in L2- and EFL-research to
explain differences in native and non-native writing, it has to be borne in mind that not all
deviations can be attributed to a transfer of L1 writing conventions. The various influences
of L1 and other factors on L2 acquisition and performance will be treated in section 2.3.1.
As the present study is concerned with the English texts of German and native English
writers, some linguistic and rhetorical differences between German and English academic
writing will be considered in the following.
Contrastive Aspects of German and English Academic Writing
The German academic register can be defined in terms of formal and rhetorical conventions
that are followed in the exchange of information between experts in academic settings
2. Academic Writing and the EFL Context
11
(Ahrens et al., 1995, p. 168f). Just as with academic English, these norms and conventions
may vary across disciplines, i.e. what is regarded appropriate in the humanities and social
sciences might not be considered ’academic’ in the natural sciences and engineering, and
in addition, there is linguistic and rhetorical variation across genres and registers, e.g.
between oral and written academic discourse.
As summarized by Clyne (1991), the German academic register is marked by:
• Agentless passives,
• Impersonal and reflexive constructions,
• Hedged performatives using modals and passive infinitives,
• A large number of nominalizations and compound nouns,
• Syntactic complexity (p.57).
These elements of an impersonal and formal writing style are rather universal to the
register of academic and scientific writing (Schwanzer, 1991), and thus can be found in
German and English academic texts as well (compare the features of academic and scientific English as discussed in section 2.1). As they are employed differently in the various
academic genres and registers as well as across disciplines, they also vary in the extent to
which they are used in German and English academic writing.
Differences between the German and the English academic register have been described
extensively by Clyne (1987, 1991). His analyzes have shown, for example, that hedging is
far more common in academic German than in English. Although modal auxiliaries such as
can, may and would provided the main hedging device in both languages, agentless passives
in the German texts exceeded those in the English texts by far. The passive infinitive (ist
zu...) also frequently occurred in the German writing while its English equivalent (is
to be...) was absent in the English texts studied by Clyne (1991). In addition to such
passive constructions, there are also some impersonal constructions to conceal and reduce
the author’s responsibility (Schwanzer, 1991). Impersonal constructions are frequent in
German and English. While English writers tend to use impersonal pronouns as in it
appears or one assumes to the same extent as impersonal phrases such as it is evident,
German writers seem to have a preference for the latter. Furthermore, Clyne (1991) found
impersonal hedging to be overrepresented in the English writing by German scholars.
With regard to discourse structure, the German academic register tends to use advance
organizers commenting on the path and organization of the paper, such as The paper
begins with. . . and The second part of the paper discusses. . . , to a much lesser extent
2. Academic Writing and the EFL Context
12
than academic English (Clyne, 1987, 1991). In addition, English academic texts seem
to feature more topic sentences than academic German, while German scholars also link
paragraphs to a previous one by using ’bridge sentences’ which are absent in the English
texts. However, both of the text-structuring devices found to be more typical for English,
i.e. advance organizers and topic sentences, were used to a much greater extent by the
German scholars in their English than in their German writing. The English texts written
by Germans thus followed the English rather than the German rules for text organization.
In general, the studies by Clyne suggest that, in the German academic register, content
is more important than the formal organization of knowledge and that academic English
is much more explicit about its structure. Linearity and symmetry are highly valued
in academic English while they are not regarded necessary in German academic writing.
German texts are marked by digressions, i.e. propositions and text segments that are not
directly related to the overarching proposition of the section but provide more general or
peripheral information on the topic (often referred to as Exkurs in German):
While digressiveness in texts by English speakers can be attributed to careless planning and is regarded as an infringement of the norm, digressive texts by German
speakers are generally planned as such. Digressions fulfill specific functions, such as
providing theory, introducing the ideological framework, entering into polemic with
other scholars or schools or giving the historical background (Clyne, 1991, p. 51).
Texts in English by Germans are even more digressive and asymmetrical than those in
German. According to Clyne (1991), this may be referred to the difficulty of writing in
a foreign language, but it also indicates that “German-speaking academics are unable to
adapt their writing patterns where the rules are not very easy to grasp and intricately
linked with the sociocultural system” (ibid, p. 56), as is the case with textual organization
in terms of symmetry and linearity.
Besides those differences between German and English academic writing, it has to be
noted that, with regard to the growing dominance of English in international scientific
communication, the conventions of academic discourse across languages and cultures are
increasingly approaching each other. So, Clyne (1991), for example, observes that “the
natural and behavioral sciences have reached a consensus (based on English discourse rules)
on international conventions for academic texts” (p.49). Likewise, the English writing
conventions might be considered as international standards in other academic disciplines.
2. Academic Writing and the EFL Context
13
2.3 English as a Foreign Language
As has been noted repeatedly in this paper, English has become the international language
of scientific research (and other branches such as international business and technology). It
is therefore of increasing importance to speakers of other languages and plays a dominant
role in second and foreign language learning and teaching.
The dominance of English in international communication is generally attributed to
historical developments. So, both the UK and US governments have promoted the use
of English as part of their foreign policy, and “in the latter part of the 20th century the
USA has taken a leading role in many academic fields of research, thus increasing academic
publishing in English at the expense of other languages” (Hoadley-Maidment and Mercer,
1998, p. 284). As a result, most scientific literature is nowadays published in English
and “countless students and academics around the world must now gain fluency in the
conventions of English-language academic discourses to understand their disciplines, to
establish their careers, or to successfully navigate their learning” (Hyland, 2006, p.24). As
English-language publications reach a larger international audience and are also cited more
often, more and more scholars from all over the world publish in English rather than in
their own language. Consequently, Hyland (2006) notes, “while research may be largely
communicated through English, it is increasingly coming from countries where English is
a foreign language” (p.26).
Differing in their social and communicative needs for learning and using the English
language, non-native speakers may be distinguished as users of EFL or ESL (Grabe and
Kaplan, 1996). The term EFL is generally referred to speakers of English living in countries
in which English is not an official language, but often is the dominating second language in
science, education and technology as in China, Indonesia and the countries of Europe. ESL
countries are those in which English either is an official language (e.g. India and Singapore),
the main language of instruction and administration (e.g. Malaysia), or the major language
of the community (e.g. Canada and the USA). Thus, ESL learners need to learn English
in order to be able to actively take part in the L2 community while the motives of EFL
learners are rather specific, such as to enhance (academic) career development.
Within the context of EFL and ESL research another distinction is inevitable, namely
that of native versus non-native English. While native English (as exhibited, for example,
in essays and other written text by native English speakers) is typically regarded as the
norm, composition and performance of non-native English speakers is generally evaluated
2. Academic Writing and the EFL Context
14
by determining the deviations from that native English standard. The term native refers
to the speakers’ first language, i.e. their mother tongue. This, however, by no means implicates that native English is uniform. Instead, language proficiency might differ among
native speakers just as with non-native speakers of English and, most probably, “not all native speakers conform to what prescriptive grammarians would see as the correct use of the
language” (Müller, 2005, p. 48). Lorenz (1999) therefore remarks that “it may not necessarily be desirable for language learners to emulate the verbal behavior of native speakers”
(p. 72), but, at the same time, he raises the issue of idiomaticity and acceptability:
Idiomaticity is a probabilistic concept; it hinges on what kinds of structures are
most likely to occur. If a given utterance complies with the basic rules of English as
codified, it may still be seen as unidiomatic - or “un-English” - by native speakers. [...]
A pattern which is exclusive to learner language is [not] necessarily not part of the
target language, or even altogether faulty. [But] acceptability judgments will always
remain a native speakers’ prerogative (ibid).
Furthermore, the use of native English as assessment criteria seems to represent a kind
of more authentic English than some idealized notion of what is to be considered Standard
English in textbooks and style manuals. In the American Heritage Dictionary of the
English Language, Standard English is defined as “the variety of English that is generally
acknowledged as the model for the speech and writing of educated speakers”3 . While British
and American English are the preferred English standard varieties to be adopted in English
language teaching (Bourne, 1998), it is primarily the American English rhetorical patterns
that are considered as the norm in contrastive rhetoric (Connor, 1996). At the same
time, however, many researchers observe an increasing tolerance towards non-native norms
of English in international academic discourse. In fact, non-native speakers outnumber
native speakers of English nearly fourfold by now and most interactions in English are
taking place among non-native speakers (Seidelhofer, 2005; Hyland, 2006). English thus
functions as a global lingua franca for individuals from different cultural and first language
backgrounds. Hyland (2006) describes English as a Lingua franca as
a variety of English which does not assume adherence to all anglo communication
conventions and where traditional native-speakerness holds no advantages. Here academic users of English are no less proficient than native speakers of that language and
3
Standard English. (2000). In The American Heritage Dictionary of the English Language, 4th ed.
Retrieved March 28, 2008, from <www.bartleby.com/61/>.
2. Academic Writing and the EFL Context
15
they are not aspiring to speak a standard English variety. What matters is clarity and
comprehensibility and L1 English speakers may need to adjust their language to new
norms of international academic communication (p.29).
Due to the extensive use of ’English as an international language’ it cannot be denied
that “English is being shaped at least as much by its nonnative speakers as by its native
speakers” (Seidelhofer, 2005, p. 339) and that non-native varieties of English (also referred
to as ’International Englishes’ or ’World Englishes’) may and should be equally represented
in eventual standards for international academic English.
2.3.1 Aspects on L2 Composition and Assessment
Writing in a second or foreign language4 is a challenging task as, in addition to the complex
cognitive processes associated with L1 composition, it involves a number of additional
variables such as social, cultural and educational factors (Grabe and Kaplan, 1996). Some
of these variables have already been approached in section 2.2 implying an influence of
the native language and culture on L2 written texts. L2 writing, however, not only differs
from native writing as far as the structure and quality of the written products is concerned,
but also with respect to the composition process itself (Silva, 1993). The role of the first
language thus has been a major concern to L2 acquisition research from its early beginnings.
Following the assumption that there was some interference between the native and the
second language, the contrastive analysis approach claimed that errors made by L2 learners
were caused by their L1, and that a greater difference between the two language systems
would account for greater difficulty and deviation in L2 performance. It was followed that
“by establishing the linguistic differences between the learner’s L1 and L2” it was possible
“to predict what problems the learner of a particular L2 would face” (Ellis, 1992, p. 7).
Empirical investigation, however, clearly indicated that structural differences between languages did not inevitably lead to difficulties and that grammatical errors could not be
attributed entirely to language transfer. Instead, it was found that “interference was more
likely to take place when there was some similarity between the first and second language
items than when there was total difference” (ibid, p. 33). The studies suggested that it
4
Though, in L2 instruction, the distinction between second and foreign language is important as regards
the social and communicative needs of their respective learners, for the discussion of L2 composition
processes, the two are used interchangeably to refer to any language that is not learned or acquired as
the first language.
2. Academic Writing and the EFL Context
16
was rather a lack of contrast between L1 and L2 that was to provoke errors5 .
Dissatisfied with the outcome of contrastive analysis, error analysis attempted to approach language transfer phenomena from another perspective. Instead of trying to predict
learner difficulties, error analysis was designed to first study the errors and mistakes in the
performance of L2 learners systematically in order to then find explanations for them. The
findings of error analysis further evolved into the study of ’interlanguage’ assuming that
the learner constructs a separate linguistic system that has features of the first and second
language as well as features distinct from the two. The interlanguage hypothesis implicates
that the utterances the L2 learner makes “are not erroneous, distorted malformations of
the utterances of native speakers, but they are systematic” and that ”the learner (primarily
unconsciously) sets up hypotheses about his L2 and tests them in communication (i.e. with
native speakers or his teacher)” (Hopkins, 1982, p. 39). The interlanguage therefore is a
system ever-developing towards greater communicative competence and, in addition, the
interlanguage approach suggests that “the interlanguages of various learners have common
features, some of which are and some of which are not traceable to L1/L2 system contrasts” (ibid, p. 46). Furthermore, the contrastive analysis hypothesis was later modified
recognizing that the L1 may influence L2 acquisition and performance in ways other than
through transfer. Thus the finding that there needed to be some similarity between L1
and L2 for interference to take place was partly attributed to the fact that learners tend
to avoid using structures that are very different from or even absent in their L1. It was
further suggested that “learners may use the L1 as a resource from which they consciously
borrow in order to improve their performance (i.e. they ’translate’)” (Ellis, 1992, p. 7).
Learner strategies like this may also lead to what is generally known as overrepresentation
and underrepresentation. As Hopkins (1982) points out, the L2 writer may “consciously
or unconsciously [underrepresent] forms and structures with which he is uncomfortable or
unfamiliar, or which have caused him trouble in the past; conversely, he may overrepresent
forms and structures of which he is sure” (p. 45). However, as he goes on, an overrepresentation of certain linguistic items “may very well be due to the overemphasis on these
forms in textbook materials and in class rather than on the presence or absence of contrast
[between the L1 and L2] alone” (ibid). Finally, Ellis (1992) concludes,
the L1 is a resource of knowledge which learners will use both consciously and
subconsciously to help them sift the L2 data in the input and to perform as best as
5
For an overview of studies investigating language transfer see, for example, Ellis (1992).
2. Academic Writing and the EFL Context
17
they can in the L2. Precisely when and how this resource is put to use depends on a
whole host of factors to do with formal and pragmatic features of the native and the
target languages (i.e. linguistic factors) on the one hand, and the learner’s stage of
development and type of language use (i.e. psycho and sociolinguistic factors) on the
other hand (p. 40).
L1 interference thus interacts, for example, with the writers’ communicative strategies,
their knowledge about the target language and their knowledge of the world. According
to Grabe and Kaplan (1996), “it is widely recognized that L1 students have some implicit
knowledge of rhetorical plans, organizational logic, and genre form in their native language
[but] it is not at all clear that students have the same implicit knowledge with respect
to L2” (p. 142). Furthermore, individual learner differences concerning age, aptitude,
motivation, personality and cognitive style (Ellis, 1992) have as much an impact on L2
composition as the form of input, i.e. whether the language is ’acquired’ by exposure in
natural settings or ’learned’ by formal instruction. L2 writing strategies further interact
with the conditions of the instructional setting and the specific teaching methods so that
the formality of the instructional situation, for example, may influence L2 learners’ text
as well. Finally, there is the impact of culture on writing conventions and reader/writer
expectations as discussed in section 2.2.
These various factors influencing foreign language writing, obviously, also have to be
taken into account in L2 instruction and assessment. EFL writing assessment and evaluation is carried out in various academic settings including such standardized tests as
TOEFL (Test of English as a Foreign Language) and MELAB (Michigan English Language Assessment Battery), but also institutional placement tests in universities and other
ESL/EFL composition programs. These tests are generally concerned with the written
products rather than with the composition process and focus on specific features that are
supposed to be a measure of writing skills. Thus the evaluation criteria established for
the TOEFL Test of Written English contain, for example, “rhetorical organization and
structure, correct sentence- and phrase-level syntax, syntactic variety and range of structures [and] detailed rhetorical support and cohesion” (Hinkel, 2002, p. 25). Similarly, the
MELAB assessment criteria include “text cohesion as demonstrated in the use of sentence
transitions and lexical connectors, coherence and the evidence of structure and planning,
the sophistication of communicated ideas [and] fully fluent, broad, and appropriate vocabulary” (ibid.). In institutional placement tests writing skills are also assessed with regard
to specific rhetorical and lexicogrammatical features that contribute to the cohesion and
2. Academic Writing and the EFL Context
18
coherence of the text (ibid., p. 26).
It is clear from that, that cohesion and coherence are considered to be major assessment
criteria for writing quality and so it is not surprising that numerous studies analyzing
non-native English writing are concerned with text organization and the use of cohesive
devices. There is much controversy, however, as to what extent cohesive markers can be
used as indicators for writing proficiency and development. Witte and Faigley (1981),
for example, found that the number of cohesive devices correlates positively with writing
development, but they at the same time point to the fact that writing quality “depends a
great deal on factors outside the text itself, factors which lie beyond the scope of cohesion
analyses” (p. 199). These factors include non-linguistic aspects and conditions accounting
for a texts’ coherence, such as the readers’ and writers’ pre-knowledge of the subject and a
discourse medium fitting to the context. Evensen (1990) reports that “marking local logical
structure in texts by connectors [...] has been found to be a sign of development in writing
abilities throughout primary school” and that “connectors have been found to be more
frequent and diversified in EFL texts produced by advanced students than in texts written
by less advanced students” (p. 171). Moreover, subordination is generally regarded to be
a feature of syntactic maturity and complexity while coordinate structures are associated
with a low writing proficiency (Biber, 1988; Biber et al., 1998). A study by Reid (1992),
however, clearly indicates that the very high proportion of coordinating conjunctions she
found in the English writing of Arabic speakers is not necessarily to be attributed to a
deficiency in writing skills but rather to a transfer from the first language. Thus, as Grabe
and Kaplan (1996) sum up, “cohesion research does not appear to be a complete answer to
understanding writing development, [but it does] provide certain useful insights” (p. 58).
2.3.2 Previous Research on Cohesion in EFL Texts
By analyzing naturally occurring data obtained from L2 writing samples, researchers
can identify what features of text, syntax, lexicon, and discourse may require special
attention in the teaching of L2 skills to enhance and focus the quality of teaching
(Hinkel, 2002, p. 13).
As a consequence, various research has been conducted in the past decades to evaluate the
writing quality of non-native English texts. The studies included analyzes of coherence
strategies concerning the arrangement of information and the macrostructure of texts but
also specific cohesive devices on surface and sentence level. Though some of the findings
2. Academic Writing and the EFL Context
19
were rather contradictory, the studies agreed on the fact that learners of English as a
foreign language often lack cohesion and coherence in their writing.
Thus, in addition to ’coherence breaks’ such as a misleading paragraph division, an
unspecified topic or a sudden and unjustified change of topic, Wikborg (1990) reported
several reference problems in Swedish student writing. Reference in fact represents an
essential means of cohesion as with the help of reference pronouns, sentences and ideas are
related to each other. In her study, however, Wikborg found that reference items lacked a
definite reference in the text or their specification was insufficient so that the reference was
ambiguous. Moreover, the distance between the cohesive items in a chain was often too
great and there was a lack of cohesive chains in general. In another study by Lee (2003),
the narratives of Chinese ESL students also displayed ambiguity in reference and a too
great distance between cohesive ties.
With regard to the number and frequency of reference pronouns, Reid (1992) found native
English speakers to use significantly fewer pronouns in expository writing than Chinese-,
Spanish- and Arabic-speaking students of English. Hinkel (2002) also distinguished between personal and demonstrative reference. In her study, personal pronouns were found
to be more frequent in the English writing of several Asian groups of non-native speakers
while demonstratives were used at higher or similar rates than in the native written texts.
In contrast, “the slot filler it was used significantly more frequently in the essays of [native
speakers] than in the essays of [non-native speakers]” (Hinkel, 2002, p. 89). Textlinguistic
research by Biber (1988), however, suggests that personal pronouns are associated with
oral and interactive discourse rather than with formal impersonal written text. Likewise,
demonstratives are discouraged in academic writing due to their ambiguity in reference
while non-referential it is a common feature in scientific texts as it is used to project distance and objectivity (Hinkel, 2002). The use of reference items in the native written texts
thus seems to be more appropriate for the register of academic writing than that in the
non-native written texts.
As explicit markers of cohesion, sentence connectors have also been a favored object of
investigation in second language research. Most findings indicate that non-native speakers
of English tend to overuse conjunctive devices and have a limited variety of connectives at
their disposal. Chinese students, in particular, show a high frequency of conjunctive devices
in their English writing in comparison to native speakers of English (see, for example, Field
and Yip, 1992; Milton and Tsang, 1993; Hinkel, 2002). EFL students from the Nordic
2. Academic Writing and the EFL Context
20
countries, in contrast, were found to use connectors sparingly just like native speakers
(Lee, 2003), and Ventola and Mauranen (1991) found Finnish academic writers to employ
even fewer connectors than native speakers of English.
Additionally, Lee (2003) compared the use of conjunctive devices in the English writing
of Nordic and Chinese students to that in their L1. While the connector density of the
Nordic students was reported to be consistent in their EFL and L1, the Chinese students
were found to use 4.5 times more connectors in their ESL than in their L1 writing. The
similar frequency of connective expressions in the English writing of Nordic students and
native English speakers thus may be attributed to the fact that the connector densities
of the respective L1’s are at a similar level. The higher density of connectors in Chinese
ESL writing in contrast to their L1, however, cannot be explained by an influence of L1
patterns, but rather “may be due to an over-emphasis of use of connectors in ESL teaching”
(ibid, p. 297).
In addition to quantitative studies on connector usage, research has also been interested
in qualitative approaches regarding the choice of conjunctive devices. As a matter of fact,
findings were strikingly different. Lee (2003), for instance, observed a limited range of
connectors in Chinese and Nordic ESL writing affirming previous conclusions that “ESL
writers lack the variety of connectors used by the native speakers” (p. 291). In contrast,
Field and Yip (1992) report a far greater range of conjunctive devices in the essays of Cantonese speakers of English as compared to a restricted connector variety used by natives.
Their study also revealed that the non-native speakers chose expressions that were rarely
used or not used at all by the native English speakers, as, for example, the additive devices
moreover, furthermore and besides. According to Field and Yip (1992), moreover and furthermore “would seem very formal and most unlikely expressions to find in student essays,
whereas [...] besides as a conjunction and not as part of a prepositional phrase [is] common
in English speech but not in writing” (ibid, p. 26). The misuse of these expressions in the
non-native essays thus suggests that students of English as a second or foreign language
are uncertain about the (in)formality of certain expressions and their appropriate use in
certain registers.
Furthermore, though displaying an underuse of connectors in general, Finnish academics
were found to overuse certain connectors at the expense of others, i.e. additive relations
were typically realized by also while however was used as the ’favorite’ adversative connector, in their English writing (Ventola and Mauranen, 1991). The researchers further
2. Academic Writing and the EFL Context
21
note that “the use of connectors by the Finnish writers seems to be infrequent, fairly locally motivated, and somewhat haphazard and monotonous” (ibid, p. 465), but also that
a skillful use of connectors, both in terms of connector frequency and connector variety,
“seems to reflect a generally high proficiency level in English” (ibid, p. 464). Finally, a
comparison of logical linkage in advanced German EFL and native English argumentative
writing conducted by Lorenz (1999) indicates that native and non-native connector usage
are approaching each other with increasing EFL proficiency.
22
3 Sentence Connectors - Functions and Uses
3.1 Terminology
Sentence connectors are words and expressions that link clauses and sentences together1 .
The connectivity between clauses is established by means of structural and semantic relationships. Sentence connectors can therefore be considered from two perspectives. On the
one hand, there is the grammatical distinction, that is, they can be classified according to
their syntactic functions (see section 3.2). On the other hand, sentence connectors can be
classified according to their semantic properties, that is, the logical relations they indicate
(see section 3.3).
Due to their various functions, sentence connectors are also referred to as ’logical connectors’ (Celce-Murcia and Larsen-Freeman, 1983), ’(logical) conjunctions’ and ’conjunctive
expressions’ (Halliday and Hasan, 1976; Hinkel, 2002), ’linking signals’ (Leech and Svartvik,
1994), or simply by the more general expressions ’connectors’ (Evensen, 1990; Ventola and
Mauranen, 1991) and ’connectives’ (Rudolph, 1996) in recent grammars and literature. In
the present paper, the terms ’sentence connectors’, ’connectors’ and ’connectives’ are used
synonymously to refer to any of the clause-linking devices examined in this study.
Obviously, clause - as opposed to single words and phrases - represents a distinguishing
feature in the description of sentence connectors. According to Biber et al. (1999), a clause
is a unit structured around a verb phrase. It is composed of a subject and a predicate,
semantically corresponding to a topic and a comment. Together, these elements express
a proposition. For the purpose of this study, clause is also defined as “grammatical unit
corresponding to one of the two parts of a connective relation” (Rudolph, 1996, p.14).
1
Although, literally, the term ’sentence connector’ seems to account for words and expressions linking
complete sentences together, for the purpose of this study it is used in a broader sense, including clauselinking expressions such as conjunctions. Phrase-level connectors introducing noun phrases, such as
because of and due to, are not included in this category.
3. Sentence Connectors - Functions and Uses
23
3.2 Structural Relations between Clauses
3.2.1 Coordination
Grammar provides several ways of linking clauses and sentences together (see, for example,
the English Grammars of Greenbaum and Quirk, 1990; Sinclair, 1990; Leech and Svartvik,
1994). First of all, clauses can be coordinated by a small set of coordinating conjunctions,
such as and, or and but. Coordination is a relationship of equality, that is, the conjoined
elements are of equal importance. Moreover, it is often “a ’looser’ connection than the
others, because it is more vague and less emphatic. It is more characteristic of informal
than of formal style” (Leech and Svartvik, 1994, p. 180). Besides clause-level coordination,
coordinating conjunctions are also used to link words and phrases. This is mostly the case
for such correlative coordinators as both...and, either...or and not only...but also. For the
study of sentence connectors, however, coordinators are only relevant if they link clauses
or sentences, as the ones in bold in the examples below (square brackets indicating an
independent clause):
(1) [A lot of different and very powerful systems have been developed.] But [these systems
normally tend to be large and expensive.] (GE96Ros)2
(2) When a voltage, V, is applied between the conductor and insulator [an electric charge is
created] and [this alters the surface free energy balance]. (BE06Her)
In (1), the second sentence is linked to the previous one by means of the sentence
connector but. The coordinating conjunction and, however, only links two adjectives in
both of the sentences. In this case, it is not considered as a sentence connector. The same
is true for the first and in (2), linking the nouns conductor and insulator. In contrast,
the second and (in bold) clearly connects two independent clauses and therefore acts as a
sentence connector.
Furthermore, when coordinated clauses share elements with a preceding clause, the element is generally not repeated in the second clause. Ellipsis, “the omission of elements
2
The following examples are excerpts from the research articles that are used as primary sources of the
comparative study presented in this paper. For better assignment, each research article has been given
a reference code specifying the English language variety (AE - (native) American English, BE - (native)
British English and GE - German EFL), the year of publication (86 - 1986, 96 - 1996 and 06 - 2006)
and the first three letters of the author’s name. The reference codes are listed with their corresponding
bibliographical references under Primary Sources.
3. Sentence Connectors - Functions and Uses
24
which are precisely recoverable from the linguistic or situational context” (Biber et al.,
1999, p.156), is very common for coordinated clauses.
(3) This converted NO to NO2 at ambient temperatures but <this> allowed unhindered passage
of NO2 . (BE86Bot)
Example (3) illustrates the ellipsis of the subject this, which can easily be recovered from
the context and has therefore been left out in the second clause (indicated here as <this>).
There are elliptical versions, however, creating much more confusion in determining the
clausal elements:
(4) The amount of the thermal flux depends not only on the distance between the tip and the
sample but also on the local thermal properties of the sample surface. (GE96Oes)
With regard to example (4), there are two ways of analyzing the coordination of the
clause constituents (cf. Greenbaum and Quirk, 1990, p.271f.). One possibility is to examine
the construction as a reduced version of clause coordination in which both the subject
(The amount of the thermal flux) and the predicate (depends) have been omitted. A
reconstruction of the sentence would then look as follows (ellipted material indicated in
<>, square brackets indicating an independent clause):
(4a) [The amount of the thermal flux depends not only on the distance between the tip and the
sample] but [<it> also <depends> on the local thermal properties of the sample surface].
Linking two independent clauses, the coordinating conjunction but would in this case
clearly represent a sentence connector.
Another possibility, however, is to examine the construction as a single clause containing
two coordinated prepositional phrases (indicated in brackets):
(4b) The amount of the thermal flux depends [not only [on the distance between the tip and the
sample] but also [on the local thermal properties of the sample surface]].
Here, the correlative coordinator not only...but also would not connect two clauses, but
simply conjoin phrasal material, not meeting the conditions for being regarded as a sentence
connector.
In the present study, equivocal sentence constructions like these are not included in the
analysis. This means, if coordinated clauses have been reduced to words and phrases by
means of ellipsis, the linking words are not considered as sentence connectors.
3. Sentence Connectors - Functions and Uses
25
3.2.2 Subordination
Another way of putting clauses together is subordination. In subordination, the clauses are
in a relationship of inequality, that is, a dependent clause is subordinated to an independent
or another dependent clause. Most subordinating conjunctions, such as if, although, because
and while, introduce an adverbial clause. The subordinate adverbial clause often contains
information which is already known or expected by the reader, while the main clause
introduces new ideas. The information in the adverbial clause also becomes less salient
than that in the main clause: “Putting an idea in a main clause is like shining a spotlight
on it; putting it in a subordinate clause [...] is like placing it in the shadow” (Leech et al.,
2006, p. 198). Note, for example, the shift in importance in the following complex sentence
(the subordinate adverbial clause is indicated in italics):
(5a) Although there is much work left to be done to rival nature, there are several significant
testing differences that account for the orders of magnitude difference in adhesion strength.
(AE06Nor)
(5b) Although there are several significant testing differences that account for the orders of magnitude difference in adhesion strength, there is much work left to be done to rival nature.
By shifting the subordinating conjunction although from one clause to the other, not
only the sentence structure has been changed in (5b), but also the emphasis of its ideas.
Subordination, therefore, seems to be more explicit and emphatic than coordination in
that the importance of a proposition can be more clearly specified.
Furthermore, subordinating conjunctions may introduce finite (6), non-finite (7) or verbless clauses (8):
(6) While this approach is appropriate for a high-volume production device, it requires many
additional masking levels and processing steps to accomplish. (AE96Spa)
(7) The pattern of the silicon dioxide platform can be modified to further enhance surface compliance while maintaining structural stability. (AE06Nor)
(8) Our filter design, while significantly different from previous reported designs, operates in this
range. (AE96Bro)
Some subordinators are bound to the use in finite clauses (e.g., as, because, in order that
and so that), while others can be used in both finite and non-finite (verbless) clauses (e.g.,
although, if, since, while and when). Their use in verbless clauses, however, must not be
3. Sentence Connectors - Functions and Uses
26
confused with so-called formulaic expressions, such as if necessary, if so or if possible. In
contrast to formulaic expressions, a verbless clause can usually be related to a finite clause
with the verb be, and it has the same subject as the main clause (Biber et al., 1999).
3.2.3 Adverbial Link
In addition to coordination and subordination, clauses can be joined together by use of
linking adverbials, such as therefore, however and nevertheless. “The primary function of
linking adverbials is to state the speaker/writer’s perception of the relationship between
two units of discourse” (Biber et al., 1999, p. 875). Thus, in contrast to function words such
as conjunctions, linking adverbials establish connections between clauses by means of their
semantic properties rather than through syntactic binding. In contemporary grammar
books, they are also referred to as linking adjuncts, sentence adjuncts (Sinclair, 1990),
sentence adverbials (Leech and Svartvik, 1994), conjuncts (Greenbaum and Quirk, 1990)
and conjunctive adjuncts (Halliday and Hasan, 1976).
Unlike conjunctions, whose position is limited to the clause boundary, adverbials are
mobile and can occupy several positions in a sentence, e.g.:
(9a) Nevertheless, work is going on to smooth out the lower shoulder of the resist profile.
(GE86MüS)
(9b) Work is going on, nevertheless, to smooth out the lower shoulder of the resist profile.
(9c) Work is going on to smooth out the lower shoulder of the resist profile, nevertheless.
Typically, linking adverbials are also orthographically separated from the rest of the
clause, that is, they are often marked off by commas (as in (9a)-(9c)). Furthermore,
they generally connect longer passages of text which themselves might contain complex
coordinate and subordinate structures:
(10) The highest values are reached [when [all the dipoles are aligned along the same direction]
and [have the same polarity]], thus contributing with the same sign to the net piezoelectric
response. (GE06Ton)
In example (10), a complex and a simple clause are joined together by the linking adverbial
thus. The first part of this connection represents itself a complex sentence consisting of an
independent clause and two coordinated subordinate clauses being linked together by the
coordinating conjunction and, and introduced by the subordinating conjunction when.
3. Sentence Connectors - Functions and Uses
27
Finally, linking adverbials can be realized by various syntactic structures. In addition to
such simple adverbs as thus, though and however, they can be composed of adverb phrases
(e.g., even so, more precisely), prepositional phrases (e.g. in addition (to), as a result (of)),
finite clauses (e.g., that is (to say)) and non-finite clauses (e.g., added to that), and often
occur in combination with conjunctions (e.g., and therefore, and/but also, but instead).
3.2.4 Overlap of Syntactic Categories
As has been shown in the previous sections, sentence connectors can be drawn from different
syntactic categories. These include coordinating and subordinating conjunctions as well
as sentence adverbials which might themselves include various word classes such as nouns,
verbs and prepositions. One difficulty in determining sentence connectors lies in the overlap
of connective expressions and other syntactic categories. Thus, several conjunctions, such
as after, as, before, but and since, can also function as prepositions, as illustrated below:
Preposition:
(11) Since the invention of the scanning probe techniques several types of probes have been employed for imaging of thermal material properties in a scanning thermal microscope (SThM).
(GE96Oes)
Conjunction:
(12) Since this filter design is self priming, it can be operated in two modes. (AE96Bro)
Moreover, there are words and expressions that belong to more than one syntactic category, both syntactic forms serving as clause-linking devices. Though, so, yet and neither,
for example, can link clauses and sentences as either a conjunction or a linking adverbial.
As a subordinator, though is closely related to although and even though introducing a
concessive circumstance clause, while as a linking adverbial it is similar to however in signaling a contrastive relationship between two propositions. “So, yet, and neither are like
coordinators in that they are fixed at the clause boundary [...] but they are like linking
adverbials in that they easily combine with coordinators (and so, and yet, but neither)”
(Biber et al., 1999, p. 80).
In addition, so seems to be a particularly problematic case, because it is also closely
related to the complex subordinator so that. As Greenbaum and Quirk (1990, p. 324)
point out, “the subordinator so is indistinguishable from the conjunct so in asyndetic
coordination, but if and is inserted so is unambiguously the conjunct” (p.324). Asyndetic
3. Sentence Connectors - Functions and Uses
28
coordination means that the coordinator (in this case and) is not present, but could be
inserted, as indicated in the following example:
(13) The oscillator frequency is in the megahertz region, <and> so the electronic control is fast
enough to regulate environmental stimulation. (GE96Ros)
Likewise controversial are the syntactic roles of for and so that, which, according to
Greenbaum and Quirk (1990), are “on the gradient between ’pure’ coordinators and ’pure’
subordinators [...] (in the meaning ’with the result that’)” (p.263).
Finally, linking adverbials often overlap with other adverbial classes. This is also the
case for thus and further, which can function as both linking and circumstance adverbials,
sometimes even implying both functions at the same time:
Linking Adverbial:
(14a) It is further possible to build an array of sensor elements that are modulated with different
frequencies. (GE86MüL)
(14b) Further, it is possible to build an array of sensor elements that are modulated with different
frequencies.
Circumstance Adverbial:
(15) The pattern of the silicon dioxide platform can be modified to further enhance surface
compliance while maintaining structural stability. (AE06Nor)
Linking and Circumstance Adverbial:
(16) Each of these different corrections is performed on an identical challenging test pattern,
thus creating a set of proximity corrected pattern sites that differ only in the values of the
parameters used for the correction process. (AE86Has)
The two uses of further (examples (14) and (15)) are easy to recognize: While the linking
adverbial can be shifted to sentence initial position (14b), the circumstance adverbial is
bound to the verb it qualifies (15). In (16), however, thus can be interpreted in two ways:
On the one hand, it is linking two clauses together by indicating that the event in the
second clause is the result of the event in the first. On the other hand, it is used as a
circumstance adverbial implying the meaning ’in this way’.
3. Sentence Connectors - Functions and Uses
29
3.3 Logical Relations between Clauses
Sentence connectors, such as conjunctions and linking adverbials, can indicate a wide range
of logical relations. By explicitly signaling the connections between clauses, sentences and
even larger parts of text, they serve as important cohesive devices. In fact, “conjunctive
elements are cohesive not in themselves but indirectly, by virtue of their specific meanings”
(Halliday and Hasan, 1976, p. 226). Though they are grammatical categories in the first
place, it is their lexical meaning that makes them signal certain relationships between the
components of a text. According to Halliday and Hasan (1976), there are four main types
of logical relation:
1. Additive: and, also, furthermore, in addition
2. Adversative: but, although, however, nevertheless
3. Causal: since, because, therefore, as a result
4. Temporal: first, second, then, finally
This distinction is rather broad and more specific meaning relations, i.e. cause, result and
purpose, are subsumed under one of the four main categories, that is in this case the causal
relation (cf. sections 3.3.1 to 3.3.4). Due to the complexity of semantic relationships, there
are numerous other ways of categorizing meaning relations. Several English Grammars
are dealing with each of the specific subrelations individually using a diversity of different
terms and classifications. Enumerative connectors (first, second, next), for example, are
treated under the category of ’ordering points’ by Sinclair (1990), ’listing and adding’ by
Leech and Svartvik (1994) and ’enumeration and addition’ by Biber et al. (1999), rather
owing to their additive character, whereas Halliday and Hasan (1976) prefer to cover them
under the category of temporal connectors as they are used to list events or steps of a
process chronologically in time. Though all these categorizations seem to be eligible, the
study in hand is based on the semantic categorizations as suggested by the latter.
Halliday and Hasan (1976) also draw a distinction between external and internal conjunction: External connectors are expressions that refer to external phenomena that language
is used to talk about, whereas internal connectors refer to relations inherent in the communication process, in terms of interaction between reader and writer. The external/internal
distinction applies to each of the types of logical relation, but is not always clear-cut. In
the present study, a differentiation between external and internal relations is only made
where relevant.
3. Sentence Connectors - Functions and Uses
30
3.3.1 Additive Relation
The additive relation is used to signal the addition of some extra information. By using
additive connectors such as and, also, besides and furthermore, “the [writer] wants [two
utterances] to be as it were added together and reacted to in their totality” (Halliday and
Hasan, 1976, p. 246). Information can also be added, however, by giving an alternative
(using the coordinating conjunction or), by indicating similarity between two propositions, or by exemplification. Similarity can be signaled by sentence connectors such as
equally, likewise and similarly. Exemplificatory expressions, such as that is, for example
or in other words, are used to add more specific information on a subject:
Addition:
(17) The ceramic material titanium diboride is primarily known for its extreme hardness and its
chemical stability even at high temperatures. Furthermore TiB2 is an excellent conductor
of electricity and heat, a property very rare among ceramic materials. (GE06Sch)
Exemplification:
(18) The use of surface micromachining circumvents the design limitations of conventional bulkmicromachined resonators. For example, the straight-forward integration of a dual resonator configuration can reduce the sensitivity of the pressure sensor to variations in both
temperature and packaging stress and electrical crosstalk. (BE96Wel)
3.3.2 Adversative Relation
The adversative relation is used to signal contrast implying the meaning ’contrary to expectation’ (Halliday and Hasan, 1976, p. 250). Contrastive sentence connectors like but,
in contrast, while and whereas basically indicate that the ideas connected are incompatible
with each other, while concessive connectors (e.g. although, still, however, nevertheless)
express a certain reservation about claims being made in the preceding clause:
Contrast:
(19) The triple layer results are approximately four times the single-layer results, whereas simple
theory would predict three times. (BE06Har)
Concession:
(20) Resonant pressure sensors consist of a resonant element attached to one or more diaphragms,
although the integration of the diaphragm and resonator into a single element is possible.
(BE96Wel)
3. Sentence Connectors - Functions and Uses
31
Moreover, adversative connectors such as instead, rather and on the contrary can signal
a correction in the sense of ’as against what has just been said’, whereas dismissive
expressions (e.g. all the same, at any rate, in any case) presuppose that “some circumstances have been referred to which are then dismissed as irrelevant - either because it
does not matter whether they obtain or not [...] or because it does not matter which of
the given set of circumstances obtains” (Halliday and Hasan, 1976, p. 254).
3.3.3 Causal Relation
The causal relation is used to establish a cause-and-effect relationship between two propositions. While causative conjunctions and adverbials, such as as, since, for, because and in
that indicate the reason for what was stated before, resultative connectors (e.g. hence,
thus, therefore, consequently, as a result) introduce effects, results and consequences
of the facts and events in the preceding clause. In addition, causal connectors can express
the logical relations of purpose (e.g. in order to, so that, for this purpose) and condition
(e.g. if, as long as, unless, otherwise). The specific meanings, however, are not always
clear-cut. So, the linking adverbial so can be equally used in the sense of ’as a result of
this’, ’for this reason’ and ’for this purpose’ (Halliday and Hasan, 1976, p. 257), and the
subordinating conjunction so that can initiate a resultative clause and a purpose clause as
well:
Effect/Result:
(21) In our case, the radius of curvature of the fluid in the exit area was several millimeters, so
that [For this reason/As a result] the pressure due to the surface tension was negligible.
(AE06Bro)
Purpose:
(22) It also needs to be reusable, so that [in order that/for the purpose that] clogging should be
minimized and reversible. (AE06Bro)
3.3.4 Temporal Relation
The temporal relation is used to express sequence in time, that is, one event is subsequent
to the other. In the case of sentence connectors like next, then, later and afterwards, the
“temporal relation is paralleled by the sequence of the sentences themselves: the second
sentence refers to a later event” (Halliday and Hasan, 1976, p. 262). Other connectors
3. Sentence Connectors - Functions and Uses
32
initiate a proposition or an event that is simultaneous or previous in time (e.g. meanwhile,
simultaneously and at the same time, and earlier, before and previously, respectively).
There are also sentence connectors that are temporally cohesive “not because [they stand]
in some particular time relation to the presupposed sentence but because [they mark] the
end of some process or series of processes” (Halliday and Hasan, 1976, p. 263). Expressions
like this are used to signal conclusion (e.g. finally, at last, eventually) and summary
(e.g. altogether, in brief, to sum up). In addition, the temporal relation can occur in a
correlative form, “with a cataphoric time expression in one sentence anticipating the
anaphoric one that is to follow” (ibid.), as, for example, first...second...third, first...then
and at first...in the end.
Finally, care must be taken with the use of temporal expressions like now, up to now, at
this point and in future. These expressions are not cohesive but deictic if used in their
external function to refer to the ’here and now’ of reality. If used internally to refer to a
particular stage of the discourse in the sense of ’here and now in the text’, however, then
they are cohesive and serve as clause-linking devices:
Deictic:
(23) Electron optical design has now improved so that beam diameters of 20 nm are possible even
for electron energies of 10 eV and below. (AE06Pea)
Cohesive:
(24) We now suggest the following model to explain our results. (BE86Eva)
3.3.5 Overlap of Semantic Categories
The four main types of logical relation are generally easy to recognize. This is especially
true for sentence connectors containing the intended meaning as a lexical component, such
as in addition, in contrast, consequently and as a result. There are connective expressions,
however, that have ’multiple semantic roles’, that is, they can express more than one
semantic relationship (Biber et al., 1999, p. 846). This is the case for some of the most
common subordinators, namely as, since, when and while. All of these conjunctions can
be used to indicate a temporal relation. In addition to this, when can have a conditional
connotation and while can be used to signal contrast or concession (in section 3.2.2, example
(7) illustrates the use of while for indicating time, whereas in (6) and (8) while is used in
its adversative meaning to signal concession). Since can also indicate reason, and, apart
from time, as is used to express reason or manner:
3. Sentence Connectors - Functions and Uses
33
Time:
(25) The flow enters in the middle of the channel and splits into two symmetric streams as [the
moment that] it goes over the electrodes. (AE06Dar)
Reason:
(26) It also gives good repeatability as [because] no bonding layer is required. (BE06Har)
Manner:
(27) A two-layer actuator was printed with gold electrodes, and polarised, as [in the manner as]
described earlier. (BE06Har)
It has to be noted here, that the subordinating conjunction as in (27) introducing an
adverbial clause of manner has no cohesive function, and is therefore not regarded as
a clause-linking device. The whole phrase ’as described earlier’, however, has a textstructuring function in that it refers to a previous part of the text.
The different uses of as, as illustrated above, are rather obvious. Sometimes, however,
the intended meanings of connective expressions with multiple semantic roles are not that
easy to distinguish:
(28) The seismic mass is set to a bias voltage of 2 V while the stator combs are virtually connected
to ground via the I–V converter stages. (GE06Reu)
In this example, it cannot be clearly determined whether the conjunction while introduces
a contrastive or a temporal clause. In fact, the clauses seem to blend semantic roles: two
processes carried out at the same time are contrasted with each other. The replacement
of while with contrastive and, otherwise, temporal connectors clearly shows that both
meaning relations are appropriate:
Contrast:
(28a) The seismic mass is set to a bias voltage of 2 V. In contrast, the stator combs are virtually
connected to ground via the I–V converter stages.
Time:
(28b) The seismic mass is set to a bias voltage of 2 V. Meanwhile, the stator combs are virtually
connected to ground via the I–V converter stages.
3. Sentence Connectors - Functions and Uses
34
3.4 Position of Sentence Connectors
Connective expressions and the position they occupy within clauses, sentences and paragraphs can reveal important information about their syntactic and semantic functions, as
well as about the register they are used in. Coordinating and subordinating conjunctions
generally occur in clause-initial position as they are syntactically restricted to the clause
boundary. Linking adverbials, in contrast, are more mobile than other syntactic categories.
They can be placed in initial, medial or final position (cf. examples (30a) to (30c)), though
some adverbials are only suitable for one of these positions. Moreover, “their positions are
determined to a larger extent by textual and pragmatic factors than the positions of other
clause elements, which are more determined by syntax” (Biber et al., 1999, p. 131).
According to a study by Biber et al. (1999), in academic prose, the overwhelming majority of linking adverbials is used in initial (50 percent) and medial (40 percent) position,
while the final position is rare (10 percent). This is not surprising as sentence connectors
“have an important role in letting us know what to expect at each step of a discourse. This
is one reason they tend to come at or near the start of each sentence in English” (Martin
and Rose, 2003, p. 128). Some researchers, however, suggest that the position of logical
connectors might also implicate differences in meaning. Celce-Murcia and Larsen-Freeman
(1983), for example, refer to a study of connector mobility conducted by Salera reporting
that the adversative connectors however, nevertheless and instead would be used:
1. Clause initially to indicate an emphatic, contrary-to-expectation expression, e.g.,
(29a) However, measurements showed that the sensor inductance is distributed around a statistic
average interval. (GE06Ric)
2. Clause medially to indicate a contrastive relationship that is not counter to expectation or one that expresses some reservation about the preceding clause, e.g.,
(29b) Measurements showed, however, that the sensor inductance is distributed around a statistic
average interval.
3. Clause finally to indicate a contrastive comment of afterthought that is less important
than what went on before, e.g.,
(29c) Measurements showed that the sensor inductance is distributed around a statistic average
interval, however.
3. Sentence Connectors - Functions and Uses
35
Similarly, they point to meaning differences depending on clause ordering, i.e. whether
the subordinate clause comes first (subordinating connector in sentence-initial position)
or whether the main clause precedes the subordinate clause (subordinating connector in
inter-clausal position), concluding that “one should look to the discourse context in order
to understand why speakers sequence the clause joined by a logical connector in the order
they do” (Celce-Murcia and Larsen-Freeman, 1983, p. 334).
Additionally, the position of sentence connectors is sometimes associated with certain
registers. Thus, the use of coordinating conjunctions in sentence-initial position is rather
unusual in formal writing, but more often occurs in informal contexts and conversation
(Sinclair, 1990; Biber et al., 1999). Coordinators can be used in the beginning of a sentence “to make the sentence seem more dramatic or forceful” (Sinclair, 1990, p. 376), but,
as Biber et al. (1999) point out, “there is a well-known prescriptive reaction against beginning an orthographic sentence with a coordinator” (p.83). Nevertheless, investigating
informal elements in academic English, Chang and Swales (1999) found out that “sentenceinitial conjunctions have already been ’legitimized’ in English scientific/academic writing”
(p.149). This seems to be especially true for the use of but, while initial and is still being
regarded as ’strange’ and ’risky’ by students and professors alike. In the end, the main
difficulty with sentence-initial conjunctions might be, (not) to know “when might be an
appropriate moment to begin a sentence with [’and’ or] ’but’ ” (Chang and Swales, 1999,
p. 162).
3.5 Sentence Connectors and Academic Writing
Sentence connectors are used for different purposes, among others, for the organization of
text, the insertion of personal remarks and additional considerations, or the emphasis of
significant points (Rudolph, 1984). Their main function, however, is to establish cohesion
by explicitly marking logical relations between propositions (cf. Halliday and Hasan, 1976)
(see also section 3.3). Rudolph (1984) points out that “certain types of text can perfectly
well do without conjunctions [...] [e.g. poems], whereas other types seem to be predestinated for these expressions [e.g. argumentative discourses]” (p.177). Academic writing
heavily relies on the explanation of causal, conditional, and other logical relationships. Sentence connectors thus can be used “to construct the logic of an argument from hypotheses
to evidence to conclusions” (Martin and Rose, 2003, p. 111) and to thereby “ ’signpost’
the path to coherence for the reader” (Lorenz, 1999, p. 56). The following excerpt from
3. Sentence Connectors - Functions and Uses
36
a scientific article clearly demonstrates how sentence connectors can support the logical
development of arguments:
(30) Conveniently, piezoelectricity could be used to provide the desired actuating or sensing capabilities. As silicon is not piezoelectric, usually, materials like quartz or lithium niobate
are applied. Their maximum operating temperatures are limited to 520 °C and 400–500 °C,
respectively. In contrast, langasite shows piezoelectrically induced bulk acoustic waves up
to at least 1050 °C. It does not exhibit any phase transition up to its melting temperature of
1473 °C and no pyroelectricity. Hence, this material is very interesting for high temperature
piezoelectric micro-electromechanical systems (pMEMS). (GE06Ans)
The excerpt starts with the sentence adverbial conveniently indicating that the upcoming information might be already known to the reader, and so the following sentence is
about materials which have been used so far for piezoelectricity. The argumentation is
supported by the causal subordinator as signaling a reason-result relation between the
subordinate and the main clause. Later on, the adversative connector in contrast not only
introduces a new idea, or in this case a new piezoelectric material (’langasite’), but due to
its contrastive meaning it also implies that this new material must be very different from
the ones previously used. After having explained the advantages of the new material, the
resultative connector hence explicitly signals that what follows is the logical conclusion of
what was stated before.
Evidently, though sentence connectors might not be mandatory for a sufficient text
comprehension, their proper use can eliminate lexical or syntactic ambiguities to a great
extent in that “text producers can exert control over how relations are recovered and set up
by receivers” (DeBeaugrande and Dressler, 1981, p. 74). As a consequence, many writing
textbooks (e.g., Raimes, 2005; Swales and Feak, 2004; Celce-Murcia and Larsen-Freeman,
1983) stress the importance of sentence connectors for creating textual cohesion and flow
of ideas, and, as a matter of fact, Biber et al. (1999) found linking adverbials, as well
as subordinators introducing non-finite clauses, to be particularly dense in the register of
academic prose:
A very important aspect of academic prose is presenting and supporting arguments.
The higher frequency of linking adverbials in academic prose not only reflects this
communicative need but also the characteristic choice of this register to mark the links
between ideas overtly, as these arguments are developed (p.880).
Hulková (2007), however, questions that sentence connectors are used in academic writing
with this notable frequency merely because of their textual functions:
3. Sentence Connectors - Functions and Uses
37
When dealing with the chapter Conclusion(s), are phrases such as to conclude, in
conclusion, to sum up, in sum and /or to summarize really important to the cohesion
of that particular piece of text, when the title itself already suggests that the author
is summing up what has been discussed above? Is this not more or less logical and
predictable even without a connective adjunct of summative meaning (p. 157f)?
The redundancy of such linking signals in mind, she suggests that ”in some cases, [...]
certain connective adjuncts only fancy up a text rather than contribute to cohesion as such”
(Hulková, 2007, p. 158) and points to an “aesthetic function” of connective expressions
when they are “selected according to conventions of ’good style’ or ’proper grammar’ “
(Biber et al., 1999, p. 43). Either way, be it as explicit cohesive devices or as markers of a
well-developed writing style, it is evident that sentence connectors have a prominent role
in academic writing, and that a lack or misuse of these expressions may lead to confusion
and misunderstanding on part of the reader.
38
4 Empirical Study
4.1 Purpose of the Study and Research Questions
The present study is to compare the use of sentence connectors in the English research
articles of German and native English writers. Previous research has yielded rather conflicting results on connector frequency and variety in non-native English writing depending
on the language and culture of the writers but also on their level of language proficiency.
As far as the German language and culture is concerned, studies on connector usage are
rare and likewise inconclusive. Clyne (1987, 1991)’s finding that German academic writing
is less explicit and less generous in its use of explicit text-structuring devices as compared
to academic English, rather suggests an underuse of connectors in German EFL texts. On
the other hand, Lorenz (1999) reported a slight overuse of causal connectors in the essays
of German advanced EFL students.
The study in hand is concerned with published academic papers of German and English
scientists. An overuse of connecting devices as found in non-native student writing (as, for
example, in the essays of Chinese (Field and Yip, 1992; Milton and Tsang, 1993; Hinkel,
2002) and German (Lorenz, 1999) EFL students), is not expected in the non-native written
RAs examined in this study for two reasons: First, scientists might have already gained
a higher proficiency level in English and might be more familiar and experienced with
the conventions of English academic writing than EFL students. This and the growing
dominance of ’English as a lingua franca’ in academic discourse (cf. section 2.3) rather
suggests a convergence of native and non-native discourse patterns. Second, academic
papers are generally edited and revised before they are published which is why they are
expected to be less deviant from native written papers than non-revised student writing.
Furthermore, many researchers agree that connector frequency and diversity is rising
with increasing language proficiency (e.g., Evensen, 1990; Ventola and Mauranen, 1991).
This could implicate that advanced non-native writers might pay more attention to connectors as explicit cohesive devices than native writers who might use connectors less
4. Empirical Study
39
consciously and therefore employ more common and less fancy connecting devices than
non-native writers. This and the fact that non-native writers do not have the same implicit knowledge of how to use connectors appropriately as native speakers, might facilitate
a misuse of connective expressions, not only in terms of frequency, but also with regard to
the choice and position of sentence connectors. Hence, the study will examine the following
research questions:
1. Do German scientific writers employ sentence connectors to the same extent in English academic writing as native English writers (connector frequency)?
2. Do German scientific writers choose the same kinds of English sentence connectors
and use the same diversity as native English writers (connector variety)?
3. Do German scientific writers employ sentence connectors in the same positions (i.e.
initial vs. non-initial) as native English writers (connector position)?
Furthermore, the above-mentioned dimensions connector frequency, variety and position
are to be considered for the years of 1986, 1996 and 2006 to answer the following question:
4. Has connector usage changed over the last 20 years in native and non-native English
academic writing?
Additionally, the native English RAs are differentiated between British and American
English in order to find out:
5. How do German scientific writers differ in their use of sentence connectors from native
British and native American English scientists, respectively?
It has to be noted at this point that the following comparison is not to draw any conclusions about writing quality or proficiency. This requires the analysis and evaluation of
various linguistic and rhetorical devices, such as the overall organization of text, thematic
progression, and also lexico-grammatical and structural errors, which is not the subject of
this paper.
4.2 Corpus
In order to compare native and non-native English academic writing, 34 research articles were selected from two scientific English-language journals, namely Microelectronic
4. Empirical Study
40
Engineering and Sensors and Actuators (A: Physical). The research article is a genre of
scientific writing. The language of the texts to be examined thus can be categorized as
formal written academic/scientific English. The journals were selected rather randomly,
but had to be
• electronically available to allow for a computer-based text analysis using a concordance program such as AntConc,
• English-language journals of international standing to ensure an ample variety of
publications by native and non-native English authors, and
• of the same scientific discipline, that is Physical Sciences and Engineering 1 , to ensure
that the RAs follow the same rhetorical conventions.
In addition, the selection of journals was limited by the RAs publication date. The
scientific articles to be examined were to be chosen from three decades, in particular from
the years of 1986, 1996 and 2006. As many internationally acknowledged journals were
only launched in the 1990s and later, and early volumes of older journals were in many
cases not electronically available, only a few journals left that also provided RAs from 1986
as electronic texts.
The research articles were chosen by a number of criteria. First, the authors had to be
native speakers of English (BE or AE), or German (a definition of native is given in section
4.3). The authors’ L1 was identified by examining the authors’ names and institutions,
and by looking at their biographical data enclosed at the end of most RAs. Since, in
many cases, several authors were quoted for one article (including contributors, supervising
tutors and cooperation partners of the persons who actually wrote the paper), and the ’real’
authors could not be clearly spotted, only those articles were taken into account in which
all the authors could be identified as native speakers of English or German, respectively.
In addition, attention was paid that the authors originated from different regions and
institutions than their compatriots, that is, the authors were selected from across their
countries (GB, USA, and Germany, respectively; see appendix A).
Second, only those RAs were chosen that had been published in the years 1986, 1996, and
2006. The selection of articles from three decades was to ensure the analysis of authentic
scientific writing over a longer period of time and to detect possible changes in connector
usage over the last 20 years.
1
This is the scientific discipline as suggested by the journals’ publisher Elsevier.
41
4. Empirical Study
Third, the RAs’ word count had to be about 2000 words each, allowing a minimum text
length of 1600 and a maximum text length of 2700 words as an exception. The statistical
data of the text corpus is summarized in table 4.1.
Table 4.1: Statistical Data of Text Corpus
Native English (NE)
Year of
AE
BE
NE in total
Publication Number Words Number Words Number Words
of RAs
of RAs
of RAs
German EFL
Number
of RAs
Words
2006
1996
1986
4
3
2
8518
6353
4160
2
3
3
4119
6559
6817
6
6
5
12637
12912
10977
6
6
5
13010
13419
10112
Total
9
19031
8
17495
17
36526
17
36541
Altogether, the corpus consists of 34 texts, of which 17 were written by native speakers of
English (9 AE and 8 BE) and another 17 by German EFL writers, and it comprises a total
of about 73,000 words. The number of words thereby only refers to the RAs’ continuous
text, that is, the single chapters and their headings. Abstracts, acknowledgments, legends,
captions, foot- or endnotes, and other additional texts have not been included in the corpus
as they have neither been considered for text analysis.
4.3 Authors
The authors of the research articles examined are native speakers of English or German in
equal shares. The term native is used to refer to the authors’ first language, being their
mother tongue. Though the writers may have acquired a second and/or third language
through formal instruction later in life, the native language is supposed to be the language
still being used predominantly by its speakers. The native English authors in this study
are either native speakers of British or American English. The two language varieties have
been chosen due to the fact that these are the two English standard varieties used as models
in English language teaching (Bourne, 1998). The native German authors are non-native
speakers of English using English as a foreign language.
Although not much data was available about the author’s age and their educational
backgrounds, it can be assumed that the majority, if not all of them, are university graduates having received at least a bachelor’s or master’s degree, such as the Master of Science,
since the RAs have been produced in academic contexts, or, if not within, then at least
4. Empirical Study
42
in cooperation with universities and research centers. Those RAs containing biographical data on the authors’ academic degrees and positions clearly reveal that most of these
writers had been PhD students, research assistants already having received their PhDs,
or professors at their respective university departments the time the RAs were published
(cf. appendix A). This, however, also indicates that the authors might differ greatly in
age and might therefore possess a rather different writing and language proficiency while
having similar educational backgrounds. Especially with regard to the non-native English
writers, it cannot be clearly specified when they have started learning English and how
proficient they are in their (presumably) second language. There is only fragmentary data
about their English language contacts, or if and how long they have been studying, working and/or living abroad in English-speaking countries. As English has been taught as the
main foreign language in Germany since the 1960s (in the New Laender since the early
1990s), it can be assumed though that the German EFL writers had at least spent some
years on English school education in addition to optional English courses at university.
4.4 Methodology
In order to compare the frequency and diversity of sentence connectors in the native and
non-native RAs, the corpus was searched for clause-linking devices from one of the following
syntactic categories:
(A) Linking Adverbials (LinkAdv),
(B) Coordinating Conjunctions (CConj),
(C) Subordinating Conjunctions (SConj)2 .
The search for sentence connectors and the subsequent frequency analysis was supported
by the concordance program AntConc3.2.1w. Words and expressions that did not fulfill
the criteria for being regarded as sentence connectors, i.e. phrase-level conjunctions linking
words and phrases rather than clauses and sentences (cf. section 3.2.1) and connective expressions from syntactic categories other than the ones mentioned above, were not included
in the frequency count. Ambiguous words had to be eliminated manually by examining
the item’s use in its context. In particular, this was the case for the coordinating conjunctions and, but and or often being used as phrase-level coordinators, and for linking
2
The abbreviations are used in the figures and tables.
4. Empirical Study
43
words and phrases with multiple syntactic functions, such as the subordinating conjunctions after, as, since and before and the linking adverbial besides which can also function
as prepositions. Moreover, it had to be examined whether the single items really had a
linking function in their contexts. Thus and further, for example, were not counted when
used as circumstance adverbials, but only in their use as linking adverbials (cf. section
3.2.4), and the subordinating conjunction if was not counted as a sentence connector when
used in formulaic expressions such as if necessary. In addition, relative pronouns such as
which and that were not included in the analysis although they are similar in function
to subordinating conjunctions in that they link two clauses together. The syntactically
ambiguous connectors so and yet, which can be equally considered as coordinators (due to
their restriction to clause-initial position) and linking adverbials (due to their occurrence
in combination with ’true’ coordinators, e.g., and so, and yet), in the present study, are
classified as linking adverbials.
The clause-level conjunctions and linking adverbials were then categorized in terms of
their semantic functions. This was done according to the four main categories of logical
relation as defined by Halliday and Hasan (1976):
Additive:
including addition (and, furthermore), alternative (or), exemplification (for example, for instance) and similarity (likewise, similarly),
Adversative: including contrast (but, instead) and concession (although, however),
Causal:
including cause/reason (because, since), result/inference (therefore,
thus), purpose (in order + to, so that) and condition (if, unless),
Temporal:
including sequence (after, then), simultaneity (when, while) and enumeration (first, second).
Since the line between the specific subcategories is not always clear-cut (i.e. several
adversative connectors can indicate both contrastive and concessive relations), semantically ambiguous items were categorized in terms of what appeared to be their primary
meaning. However, the corpus also included sentence connectors with multiple semantic
roles indicating more than one of the four main meaning relationships (cf. section 3.3.5).
For this reason, a distinction was drawn between causal as1 (’because’) and temporal as2
(’at the same time that’), adversative while1 (’whereas’/’although’) and temporal while2
(’during/at the same time that’), and inferential then1 (’in that case’/’as a consequence’)
4. Empirical Study
44
and temporal/sequential then2 (’after that’) (cf. table B.1). The as-construction (so) as
+ to-infinitive indicating result or purpose was treated as an extra item under the category of causal connectors. The subordinating conjunction since was only used in its causal
meaning so that no distinction was necessary for that.
Another difficulty was the classification of items indicating two or more semantic relationships at the same time. In fact, semantic concepts such as reason and time or time and
condition are closely interconnected and often occur in combination. Thus the temporal
subordinator when and the temporal linking adverbial then can both have a conditional
connotation in the sense of ’in that case’ similar to the subordinating conjunction if.
(31) This process is short and stops when [at the time/in the case that] these charges are neutralized at the electrodes. (AE06Dar)
In the above example, the subordinator when implies both, the specific point of time at
which ’these charges are neutralized’ and the more generalized time whenever the condition
that ’these charges are neutralized’ is met. The second interpretation is clearly related to
the semantic concept of condition. As, in the case of when, temporal and conditional
meanings are too closely intertwined, it was only treated under the temporal relation (the
conditional connotation still being recognized). Other semantically ambiguous items were
treated under those categories that appeared to be their primary meaning (as in the case of
as, while and then). The items that were recognized as sentence connectors are summarized
in table B.1.
For the text analysis, the number and position of the sentence connectors was noted for
each RA and arranged according to the RA’s year of publication (1986, 1996 and 2006) for
native English and German EFL, respectively. Furthermore, the connector frequencies of
the native English RAs were listed separately for American and British English to allow for
an individual consideration of these two English varieties. The data was then normalized
to a basis per 10,000 words so that the values were directly comparable.
Additionally, a statistical significance test was performed by determining χ2 for some individual and overall connector frequencies. Statistical significance testing, however, seemed
not very revealing for the purpose of the present comparison. Not only is the practical value
of statistical significance tests highly disputed in general (cf. Armstrong, 2007; Goodwin,
2007), but the question whether certain differences are statistically significant is largely
dependent on the number of words the individual frequencies are normalized to. For the
present comparison, this means that the χ2 -test reveals different significances depending
45
4. Empirical Study
on whether the frequencies are normalized to a basis per, for example, 1,000, 10,000 or
100,000 words while the respective ratios of the native English and German EFL frequencies remain the same (the variability of the χ2 -value is illustrated for the standardized
overall connector frequency in native English and German EFL in table B.2). By using a
sufficiently large corpus size, even small differences thus can become statistically significant though they may be insignificant in practice. This makes the values for χ2 appear to
be rather meaningless so that no claims of statistical significance are being made for the
results of this study.
4.5 Results
4.5.1 Overall Frequency
Table 4.2 summarizes the overall frequency of sentence connectors in the native and nonnative English RAs. In total, the German EFL writers employed fewer sentence connectors
than the native English writers with about 217 and 249, respectively. This result, however,
must be reconsidered with respect to the native English varieties: The overall frequency
of connectors in the American English RAs is relatively close to that in the German EFL
texts with about 225, while a total of about 276 occurrences per 10,000 words in the British
English RAs exceeds those in the American and non-native English RAs by far.
Table 4.2: Overall Connector Frequency (items per 10,000 words)
Year
AE
BE
Native English
German EFL
2006
1996
1986
224.23
242.41
199.52
301.04
259.19
275.78
249.27
250.93
246.88
230.59
179.60
250.20
Total
224.90
275.51
249.14
217.29
The temporal consideration of the research articles is less revealing, though it suggests
that native English connector usage has been more consistent over the years of 1986, 1996
and 2006 than non-native connector usage in terms of frequency. The assumption that
native and non-native connector usage would be converging could not be affirmed. As
table 4.2 indicates, in 1986 the frequency rates of the native and non-native RAs were
almost the same. In the articles published just recently (2006), the non-native speakers
employed slightly fewer sentence connectors than the native writers. The greatest difference
was found in those RAs published in 1996 showing a substantial underuse of connectors
46
4. Empirical Study
by German writers.
Table 4.3 lists the connectors most frequently used in the corpus. As expected, the
clause-level coordinator and was by far the most common connector in both native and
non-native texts. The clause-level conjunction but, the subordinators if and in order + to
as well as the linking adverbials also, thus and therefore were also used with considerable
frequency in all the RAs. Differences in the frequency of the single connectors are discussed
in sections 4.5.3 and 4.5.4.
Table 4.3: Most Frequently Used Connectors (items per 10,000 words)
Native English
and
also
then2
when
but
however
thus
if
in order + to
so that
therefore
as1
as2
since
although
79.67
13.96
11.22
10.95
10.40
10.40
7.39
7.39
6.84
6.84
6.02
5.75
5.75
4.65
4.11
German EFL
and
if
also
but
therefore
in order + to
since
thus
when
then1
because
for example / e.g.
after
however
while1
56.10
16.69
16.15
11.49
10.67
6.84
6.29
5.75
5.75
5.47
5.20
4.65
4.38
4.10
3.83
4.5.2 Overall Variety
With regard to connector variety, the German EFL writers employed slightly more different
connective expressions than the native English writers (61 and 57, respectively, out of 69
connectors recognized in total). The greater variety in non-native writing was consistently
found in all years investigated. Altogether, differences in the choice of connectors became
apparent as 20 out of the 69 expressions found in total were exclusively used by either
native or non-native writers (in particular, 8 by native and 12 by German writers). Of
the 8 connectors that only occurred in the native RAs, 4 were exclusively used by British
writers (i.e. similarly, conversely, thirdly and nonetheless) and one by American scientists
(even though). The remaining three (next, unless, in that) were used by both British and
American writers. Moreover, there were 16 connectors that occurred in the German and
47
4. Empirical Study
either in the British or American RAs. 10 of these expressions were found to be used by
American writers and only 6 by British writers. This might suggest that, in their choice
of connectors, the German scientists rather share the preferences of the American writers.
In the following, differences in the choice of connectors will be discussed in more detail.
4.5.3 Syntactic Categories
Figure 4.1 displays the percentage distribution of the sentence connectors found in the
native and non-native RAs across the three syntactic categories: linking adverbials, coordinating conjunctions and subordinating conjunctions. While native English writers for
the most part used coordinating conjunctions (37.36 %), the greatest number of sentence
connectors in the German EFL writing was made up by linking adverbials (38.92 %).
Subordinating conjunctions were the least used syntactic category in the native and nonnative RAs with 28.68 % and 29.22 % respectively. It should be noted, however, that the
frequency distribution of the American writers much more closely resembled that of the
German EFL writers, and that the differences in the distribution across syntactic categories
must be attributed primarily to differences between the German EFL and the British RAs
(for more detail see appendix B, table B.3).
Figure 4.1: Distribution across Syntactic Categories
Furthermore, linking adverbials constituted the greatest part as regards connector variety. Thus, from 57 different connective expressions found in the native English RAs, 34
were linking adverbials (59.7 %). In the non-native RAs the share was even higher with
48
4. Empirical Study
40 linking adverbials out of 61 different connectors found in total (65.6 %). In addition,
the native English writers used sligthly more different subordinating conjunctions (cf. table 4.4). In the following section, the frequency distribution of coordinating conjunctions
is presented in detail. Subordinators and linking adverbials are considered under their
respective semantic categories (cf. section 4.5.4).
Table 4.4: Connector Variety according to Syntactic Categories
Native English
German EFL
LinkAdv
CConj
SConj
34
3
20
40
3
18
Total
57
61
Coordinating Conjunctions
As corpus studies have shown, and is by far the most common coordinator in oral and
written discourse alike. In academic writing, however, it is more typically used as a phraselevel connector than as a clause-level connector (Biber et al., 1999). This is also reflected
in the present study.
Table 4.5: Clause-level Coordinators (items per 10,000 words)
Native English
Clause-level All Occurrences
German EFL
Clause-level All Occurrences
and
Initial
but
Initial
or
Initial
79.67
0.27
10.40
3.01
-
239.01
12.87
29.84
-
56.10
11.49
3.28
1.64
-
241.92
15.87
22.17
-
Total
Initial
93.08
0.27
281.72
-
69.24
3.28
279.96
-
Table 4.5 displays the frequency distribution of clause-level coordinators in addition to
their total number of occurrences (comprising phrase- and clause-level coordinators) and
the number of occurrences in sentence-initial position. And was the most frequently used
coordinating conjunction in both native and non-native writing, but it was more frequently
used as a clause-level connector by the native English writers. Thus, in the native RAs,
clause-level and’s made up 33 % of the total occurrences (79.67 of 239.01) complying with
49
4. Empirical Study
the findings of Biber et al. (1999) who reported a percentage use of about 35 % of and
as clause-level connector in academic prose (p. 81). In contrast, only 23 % of and’s
found in the German EFL writing were identified as clause-level conjunctions (56.10 of
241.92 occurrences in total). But and or were used with similar frequency as clause-level
conjunctions by native and non-native writers.
Moreover, more than one third of the but’s in the German RAs occurred in sentenceinitial position (3.28 of 11.49 in total) whereas in the native English RAs one instance of
initial and was found (0.27). This result, although complying with the fact that but is more
likely to occur in the beginning of a sentence than other coordinators (Biber et al., 1999),
is rather striking, as, in general, sentence-initial coordinators are considered too informal
for the register of academic prose (cf. section 3.4).
Table 4.6: Coordinators and Linking Adverbials in Combination (items per 10,000 words)
Native English
German EFL
and also
and hence
and so
and then
and thus
and therefore
and finally
but also
but instead
but on the other hand
0.82
0.82
2.46
2.19
1.64
2.19
0.27
1.10
-
0.55
0.27
0.55
1.09
1.64
0.27
0.82
0.27
0.27
Total
11.50
5.75
Furthermore, coordinating conjunctions are also often used in combination with linking adverbials. Table 4.6 shows the combinations of clause-level coordinators and linking
adverbials and their frequencies as found in the corpus (these also include combinations
in which the coordinator is not directly followed by the linking adverbial, e.g. ’and will
therefore have’ or ’and is then used to’, but disregard linking adverbials occurring with
phrase-level coordinators). The table clearly indicates that the native English writers used
clause-level coordinators twice as often in combination with linking adverbials as the German EFL writers. And so, and then and and therefore were the most frequent combinations
in the native RAs whereas in the non-native texts only and thus and and then were used
with some frequency.
50
4. Empirical Study
4.5.4 Semantic Categories
Figure 4.2 displays the percentage distribution of the sentence connectors found in the
native and non-native RAs across semantic categories. Additive and causal relations are
generally frequent in academic writing (Biber et al., 1999). They therefore constituted
the greatest part of semantic relations in the present corpus, too. While additive and
adversative connectors were used to the same extent by native and non-native writers
(about 44.2 % and 12.2 % in the native, and 42.2 % and 12.8 % in the German EFL
writing), the non-native RAs displayed a higher percentage of causal connetors (33 % as
compared to 25.6 % in native English). In contrast, temporal connectors made up 18 % of
total connectors in the native, but only 12 % in the non-native RAs.
Figure 4.2: Distribution across Semantic Categories
Additive Connectors
The frequency analysis yielded that additive connectors were more frequently used in the
native English RAs with about 110 occurrences per 10,000 words in contrast to about
92 in the German EFL texts (cf. table B.6). These total numbers, however, include the
frequencies of both additive conjunctions and additive linking adverbials and, as shown in
table 4.5, the greater number of additive connectors in the native writing clearly results
from the significantly greater use of clause-level and in the native English RAs.
Leaving the additive coordinators and and or (cf. section 4.5.3) aside, table 4.7 clearly
51
4. Empirical Study
Table 4.7: Additive Linking Adverbials (items per 10,000 words)
Native English
German EFL
Addition
additionally
also
as well
besides
further
furthermore
in addition
indeed
in fact
moreover
too
1.92
13.96
0.27
0.55
0.82
3.01
0.55
0.27
0.55
0.27
1.09
16.15
1.09
0.27
0.55
3.83
0.82
0.27
0.27
0.55
2.19
Exemplification
for example /
e.g.
for instance
that is /
i.e.
1.92
0.82
0.27
0.55
1.37
1.09
3.56
0.27
1.64
Similarity
likewise
similarly
0.27
0.27
-
Total Additives
27.38
33.93
indicates a higher frequency of additive linking adverbials in the non-native writing (33.93
as compared to 27.38 in the native RAs). The native English writers preferably introduced
additional information by linking adverbials containing the additive relation as a lexical
component, i.e. additionally and in addition, whereas the German EFL writers more frequently employed the synonymous sentence connector furthermore and the more informal
sentence-final adverbials too and as well. Sentence connectors indicating similarity were
scarcely used in the whole corpus whereas exemplificatory connectors were frequent in both
native and non-native writing.
Exemplificatory linking adverbials are common in academic prose as they are used to
clarify previous statements by adding examples and more detailed information on a subject
(Biber et al., 1999). For the appositional connectors for example and that is there also exist
the Latin abbreviations e.g. and i.e. which are mainly used in formal written texts. E.g.,
however, is more typically used to give specific examples of technical terms rather than
to introduce main ideas, and it rarely occurs sentence-initially as for for example and for
52
4. Empirical Study
instance (ibid). In the present corpus, the German EFL writers showed a clear preference
for the abbreviated versions (3.56 and 1.64 occurrences of e.g. and i.e., respectively, as
compared to 1.09 occurrences of for example and no instance of that is at all) whereas
native speakers of English made use of both full and abbreviated spelling variants (see
table 4.7).
Adversative Connectors
There were a number of adversative connectors that were only used by the German writers,
i.e., the linking adverbials nevertheless, instead, on the contrary, on the other hand and
yet as well as the subordinating conjunction though (see table 4.8). The fact that these
expressions were present in the non-native, but absent in the native English RAs, is rather
remarkable since three of the connectors, i.e. yet, nevertheless and on the other hand, were
reported among the five most common linking adverbials to indicate contrast/concession
in English academic prose by Biber et al. (1999). The native writers generally employed
Table 4.8: Adversative Connectors (items per 10,000 words)
Native English
German EFL
Contrast
but a
conversely
in contrast
instead
on the contrary
on the other hand
whereas
while 1a
yet
10.40
0.27
0.27
1.10
1.97
-
8.21
0.27
0.27
0.55
1.09
3.01
3.00
0.27
Total
14.01
16.67
Concession
although
but b (initial)
even though
however
nevertheless
nonetheless
though
while 1b
4.11
0.27
10.40
0.27
1.37
1.37
3.28
4.10
1.09
0.55
0.82
Total
16.42
11.21
Total Adversatives
30.43
27.88
53
4. Empirical Study
a smaller range of adversative connectors, but they used some of them with considerable greater frequency (as with although and however) and so the overall frequency of
adversatives was nearly the same in the native and non-native RAs (about 30.5 and 28
respectively).
The analysis of adversative connectors further revealed differences with regard to the
subcategories of contrast and concession. Concession clauses are notably common in the
register of academic writing as they are used to counterbalance facts, claims and positions
and thus provide for an objective argumentation (Biber et al., 1999). They are also “somewhat sophisticated due to the complexity of the notion of concession in written text and
text cohesion” (Hinkel, 2002, p. 138). There are a number of adversatives that can indicate
both contrastive and concessive relations, i.e. while and however, as well as the conjunction
but indicating concession rather than contrast when used sentence-initially (see section 5.4
for a more detailed discussion on initial but). In the present corpus, concession clauses
were much more frequent in the native than in the non-native texts (16.42 as compared
to 11.21, respectively) being introduced primarily with however which Biber et al. (1999)
found to be the most common linking adverbial to indicate contrast/concession in academic
texts. In the German EFL writing, concession clauses were also frequently introduced by
sentence-initial but which was absent in the native English RAs. In addition, while was
used primarily in its contrastive sense by the German writers (3 occurrences per 10,000
words as compared to 0.82 for concessive while) whereas in the native English RAs it was
equally used to indicate contrast and concession (1.97 and 1.37, respectively).
Table 4.9: Selected Adversative Subordinators (items per 10,000 words)
although
even though
though
whereas
while1
AE
BE
Native English
German EFL
1.05
0.53
0.53
6.31
7.43
1.71
-
4.11
0.27
1.10
3.29
1.37
0.55
3.01
3.83
Table 4.9 displays the frequency distribution of five selected adversative subordinators
including their individual frequencies in the native English varieties (AE and BE). The
concessive subordinators though, although and even though are synonymous in meaning,
but although is much more typical in academic writing as it is the more formal variant
(Biber et al., 1999) while even though is somewhat more emphatic (Leech and Svartvik,
4. Empirical Study
54
1994). In the present corpus, although was used more than twice as frequently in the native
than in the non-native RAs. Considering the native English varieties in detail, however,
the data clearly shows that it was almost exclusively used by the British writers (7.43)
while it was employed rather sparingly in both AE (1.05) and German EFL (1.37). The
more informal variant though was only used in the non-native RAs. The use of whereas
and while was also found to be strikingly different in the native English varieties. Thus,
the British writers seemed to prefer the former with no instances of contrastive/concessive
while, whereas the American writers showed a clear preference for the latter with a ratio
of 6.31.
Causal Connectors
The broad category of causal relations can be subdivided into several subcategories, namely
cause/reason, result/inference, purpose, and condition. In the native English RAs, the
cause/reason relationship was preferably indicated by as and since, while in the non-native
RAs since and because were used most frequently. According to Biber et al. (1999), because
is the most common reason subordinator not only in academic writing, and it is the most
clear choice because, in contrast to as and since, it refers to reason unambiguously. Causal
as and since are more typically found in formal writing than in other registers, but since is
more frequently used to indicate reason than as. Investigating causal connectors in British
English and German EFL, Lorenz (1999) identified because as ’markedly non-native’ and
in that as ’a markedly native usage’, the latter occurring in more mature writing. As table
4.10 indicates, in the present corpus, because occurred twice as often in the German RAs
while as occurred three times more often in the native texts. In that was exclusively used
by native English writers. This points to a markedly native versus non-native style in
the present corpus, too. In addition, the analysis revealed a preference for certain causal
connectors by British and American writers. Thus, causal as was employed twice as often in
BE than in AE (8.00 and 3.68, respectively) whereas in AE reason clauses were preferably
introduced by since (6.83 as compared to 2.29 in BE; cf. table B.8).
The most common linking adverbials for result and inference in English academic prose
are, according to Biber et al. (1999), thus, therefore and hence which, in most cases, are
interchangeable. They also found that “most academic texts show a clear preference for
either thus or therefore, usually using one choice at least twice as often as the other” (ibid,
p. 889). In the present corpus, this was the case for the German writers employing therefore
55
4. Empirical Study
Table 4.10: Causal Connectors (items per 10,000 words)
Native English
German EFL
Cause/Reason
as1
because
in that
since
5.75
2.74
1.37
4.65
1.92
5.20
6.29
Result/Inference
as a result
consequently
for this/these reason/s
hence
so
then1
therefore
thus
0.82
0.55
0.27
2.19
4.11
3.56
6.02
7.39
0.27
0.27
0.55
2.19
3.28
5.47
10.67
5.75
Purpose
in order + to
so as + to
so that
6.84
0.82
6.84
6.84
0.55
2.46
Condition
as long as
if
otherwise
unless
whether
0.27
7.39
0.55
1.64
0.27
16.69
1.92
1.09
Total Causals
63.79
71.70
almost twice as often as thus (10.67 and 5.75, respectively) and four times as often as hence
(2.19). The native English writers used both thus and therefore with notable frequency.
However, while thus was equally used by the British and American writers (8.57 and 6.31,
respectively), therefore was predominantly found in the British RAs (10.86 as compared
to 1.58 in AE; cf. table B.8).
Purpose clauses are also considerably common in academic texts as they are used to
“present procedures or mark recommendations for improving conditions” (Biber et al.,
1999, p. 824). Table 4.10 clearly shows that, in the present study, purpose was frequently
indicated by in order + to in the native and non-native RAs (6.86 occurrences per 10,000
words each) while the native writers also made frequent use of so that (6.84 as compared
to 2.46 in German EFL). So as + to was not very common in the whole corpus.
4. Empirical Study
56
Conditional clauses are typically introduced with the subordinating conjunction if. In
the present corpus, if was used with considerable frequency, too, though it was used more
than twice as often in the German EFL as in the native English RAs (16.69 as compared
to 7.39). To introduce a negative condition, the German EFL writers preferably used the
linking adverbial otherwise while the native speakers rather made use of the conjunction
unless. In addition, if also occurred a number of times in correlation with then1 introducing
the consequence of the if -condition. In the native English RAs the frequency ratio for
if...then was 1.37, and in the German EFL writing 1.09 in addition to 0.55 occurrences of
when...then.
Temporal Connectors
Time subordinators such as after and before are more commonly found in less formal
registers, such as fiction and news (Biber et al., 1999). In academic writing they are
occasionally employed with non-finite clauses to describe, for example, technical methods
and procedures. Enumerative and sequential adverbials such as first, second, then and
next are much more typical in academic prose since, in their external function, they are
often employed to explain processing steps in chronological order, and, in their internal
function, they are used to structure the plenitude of information given in academic texts.
As shown in table 4.11, in the present study, temporal connectors were much more frequent
in the native than in the non-native RAs with about 45 and 26 occurrences per 10,000
words, respectively. Then, as and when, in particular, were used with considerable greater
frequency by native speakers of English. Enumeration was preferably realized by first and
secondly in the native English RAs whereas the German writers made also use of related
expressions such as firstly and first of all.
Connectors with Multiple Semantic Roles
Table 4.12 shows the frequency distribution of sentence connectors that can indicate more
than one meaning relation. In the German EFL writing, connectors with multiple semantic
roles were preferably used in their non-temporal meaning, whereas in the native English
RAs then was used three times more often as temporal (i.e. sequential ’after that’) than
as causal connector (i.e. resultive ’as a consequence’/conditional ’in that case’). The
subordinator since was used exclusively in its causal meaning by both native and nonnative writers which is rather typical for the register of academic prose (Biber et al., 1999).
57
4. Empirical Study
Table 4.11: Temporal Connectors (items per 10,000 words)
Native English
German EFL
Sequence
after
afterwards
before
finally
next
now
then2
3.01
1.92
2.46
0.55
1.10
11.22
4.38
0.82
1.37
1.37
3.01
2.46
Simultaneity
as2
when
while2
5.75
10.95
2.19
0.27
5.75
1.64
Enumeration
first
firstly
first of all
second
secondly
thirdly
4.11
0.27
1.10
0.27
2.74
0.82
0.55
0.27
0.55
-
Total Temporals
44.90
26.00
Table 4.12: Connectors with Multiple Semantic Roles (items per 10,000 words)
as (causal)
as (temporal)
since (causal)
since (temporal)
then (causal)
then (temporal)
while (adversative)
while (temporal)
Native English
German EFL
5.75
5.75
4.65
3.56
11.22
3.29
2.19
1.92
0.27
6.29
5.47
2.46
3.83
1.64
The subordinator as was generally frequent in the native writing in both its causal and
its temporal meaning, while in the German EFL texts it was more frequently used as a
causal connector. The subordinator while, according to Biber et al. (1999), is generally
used as a temporal connector in conversation whereas, in academic texts, over 80 % of
its occurrences mark contrast and concession. In the present corpus, however, it also
frequently occurred as a temporal connector. In particular, temporal while made up 40 %
58
4. Empirical Study
of the total occurrences in the native and 30 % in the non-native RAs.
4.5.5 Connector Position
A notable difference between native and non-native connector usage was found with regard
to connector position. Table 4.13 clearly shows that the German writers used one third
more connectors in sentence-initial position (as opposed to non-initial position) than their
native English counterparts. About 29 percent of the connectors in the non-native RAs
occurred in the beginning of the sentence whereas the same was true for only about 19
percent of the connectors employed in the native English RAs. The overuse of sentenceinitial connectors by the German writers thereby equally applied to linking adverbials
and conjunctions, and was most notably apparent in the semantic categories of causal
and temporal connectors. However, compared to the figures given by Biber et al. (1999),
reporting 50 % of linking adverbials in academic prose to occur in initial position (p.891),
the percentages of 34.95 % in the native and 42.43 % in the non-native RAs seem to
be rather below average. The temporal consideration further suggests a rising tendency
towards sentence-initial position in both native and non-native writing, the increase in the
non-native texts being much more substantial, however (cf. table 4.13, Year of Publication).
Table 4.13: Initial Connector Usage across Categories (% of total occurrences)
Native English
LinkAdv
CConj
SConj
German EFL
Syntactic Categories
34.95 %
42.53 %
0.29 %
4.72 %
24.90 %
37.50 %
Semantic Categories
Additive
8.71 %
Adversative
37.84 %
Causal
24.46 %
Temporal
24.39 %
2006
1996
1986
Total
9.55 %
41.18 %
43.51 %
44.21 %
Year of Publication
20.32 %
35.33 %
20.06 %
26.56 %
16.61 %
23.72 %
19.12 %
28.97 %
Figure 4.3 contrasts the sentence connectors that were most deviant among native and
non-native writers in terms of position (connectors showing more than 25 % discrepancy
59
4. Empirical Study
Figure 4.3: Sentence-Initial Connectors
in the percentage to which they were employed in sentence-initial position, in descending
order of discrepancy). Again, it becomes obvious that the German EFL writers have
a clear preference for the sentence-initial use of connectors whereas native speakers of
English tend to employ linking words and phrases inbetween sentences. Only two out of
the 12 connectors listed were more frequently used sentence-initially by the native writers,
i.e. for example and adversative while. Sentence-initial subordinating conjunctions were
equally overrepresented in the non-native writing as linking adverbials, whereas causal
linking adverbials in particular were preferably used in initial position by the German
writers. Thus, in the German EFL texts, 6 out of 12 occurrences of so were found in
sentence-initial position whereas this was the case for only one out of 15 occurrences in the
native RAs. According to Lorenz (1999), initial so is less formal and “clearly a dispreferred
feature in more mature argumentative writing” (p. 63). It therefore seems self-evident that
it was preferably used inter-clausal, i.e. in combination with and, by native writers (cf.
also table 4.6).
60
5 Summary of Findings and Discussion
5.1 Main Findings
The corpus analysis has yielded the following results in answer to the research questions
(cf. section 4.1):
1. Connector frequency: German scientists tend to employ fewer sentence connectors in
English academic writing than native English writers.
2. Connector variety: German scientists tend to employ a greater variety of sentence
connectors including connectors more typical of informal registers and less mature
writing.
3. Connector position: German EFL writers use sentence connectors with significantly
greater frequency in sentence-initial position than their native English counterparts.
The results of the temporal analysis can be summarized as follows:
4. Over the years of 1986, 1996 and 2006, connector frequency has been fairly constant in
native but somewhat inconstant in non-native writing. Differences in the choice and
variety of connectors have been rather consistent in all these years, while the overuse
of sentence-initial position has increased dramatically in the German writing. A
convergence of native and non-native connector usage was not observed.
The differentiation between BE and AE has yielded another important finding:
5. German connector usage is more different from British than American English in
terms of frequency. In addition, German scientists tend to share the same preferences
for certain connectors as American scientists.
With reference to this last point, it has to be noted that there are substantial differences
between British and American connector usage, and this has in fact relativized some of
5. Summary of Findings and Discussion
61
the study’s findings. In the following, the results of the three main dimensions connector
frequency, variety and position are discussed in more detail. A reconsideration as to the
choice and frequency of sentence connectors by British and American writers, respectively,
is given in the respective sections.
5.2 Connector Frequency: A Pointer to L1 Interference?
The result that, in total, the German EFL writers employed fewer sentence connectors than
the native English writers (about 217 versus 249 per 10,000 words, respectively) points to
an influence of L1 writing patterns. German academic writers have been found to prefer a
more implicit writing style emphasizing content over form and formal organization whereas
English writers tend to be more explicit in their writing (cf. Clyne, 1987, 1991, see also
section 2.2). Sentence connectors are markers of explicitness in that they are used to
explicitly signal logical relations. The underuse of connectors in the German RAs thus
might be a result of interference from the German mother tongue. The difference in
connector frequency, however, was not substantial so as to consider whether the differences
rather account for individual writing preferences of the authors. Moreover, an underuse of
connectors by German writers was primarily observed in comparison with British English
whereas the overall frequencies of connectors in the German and American English RAs
were nearly the same. This could indicate that the German writers were aware of the fact
that discourse patterns differ across cultures and that they therefore tried to conform, in
this case, apparently, to American writing conventions rather than adhering to German
discourse patterns in their English writing (cf. the findings of Clyne, 1987, 1991, comparing
English and German texts by German writers).
5.3 Connector Variety: Native and Non-Native Preferences
In terms of connector variety, it was found that non-native connector usage was more
diversified than that of native English writers. Though the overall difference was not
substantial, it was consistently found in all years investigated. For example, in the nonnative RAs, enumerations were introduced by a whole range of possibilities, i.e. first,
firstly, and first of all. In the native RAs they were exclusively initiated by first, the
simplest of these possibilities. Adversative relations were preferably expressed by however
in the native texts while the German writers showed much more variation including several
5. Summary of Findings and Discussion
62
adversative connectors that were absent in the native texts. Native writers thus seemed
to have a clear preference for specific connectors whereas the choice of the German writers
seems to have taken place more haphazardly.
5.3.1 ’Lexical Teddy Bears’
In some cases the non-native writers also relied on so-called ’lexical teddy bears’, i.e. words
and expressions that the writers are especially familiar with and, for this reason, used as an
’all-purpose wild-card’ (Lorenz, 1999)1 . Reason clauses, for instance, were twice as often
introduced by because in the non-native RAs, whereas the native writers clearly preferred
connectors such as as and in that. Other examples for ’lexical teddy bears’ in the nonnative texts are furthermore and therefore. The overrepresentation of these items in fact
points to a well-known learner strategy, that is, the avoidance of forms and structures
learners feel uncomfortable or unfamiliar with (cf. section 2.3.1).
For example, because is the only reason subordinator that refers to reason unambiguously (Biber et al., 1999). The reason subordinators as and since both are lexically and
syntactically ambiguous in that they can also refer to meaning relations other than reason
and are equally used as prepositions or adverbs. As and in that also have been found
to occur in more mature writing and are more typical of written registers while because
clauses are more frequently used in speech than writing. In addition, therefore seems to be
a ’simpler’ choice to introduce a consequence than thus, for instance, which can indicate
a wide range of meaning relations. This suggests that the German scientific writers felt
more comfortable with the use of less ambiguous, but also less mature expressions for these
are less susceptible to misuse. However, the overrepresentation of certain connectors could
also be due to an overemphasis on these items in L2 teaching materials. It would therefore
be interesting to examine this in a future study.
5.3.2 Inconsistency in Register
Moreover, the German EFL writers used expressions more typical of informal speech than
academic writing (note, for example, the informal use of so and but in sentence-initial
1
The term ’lexical teddy bears’ was coined by Angela Hasselgren to describe high-frequency words and
phrases that are heavily overused by learners due to the fact that they are particularly familiar and
easy to memorize. Lorenz (1999) refers this phenomenon to what he terms a ’wild-card use’: “If a
given linguistic element is used as an all-purpose wild card, that usage is bound to include a number of
instances of over-extension” (p. 60).
5. Summary of Findings and Discussion
63
position, and the use of rather informal connectors like besides, too and as well). These
expressions were scarce or even absent in the native RAs. The use of informal linguistic
devices in formal written texts is a phenomenon often observed in second language research
and it is generally attributed to a lack of sensitivity to register: “One of the things that
marks speakers as non-native is their producing a logical connector of an inappropriate
register” (Celce-Murcia and Larsen-Freeman, 1983, p. 323). This does not necessarily
imply, however, that the non-native academic papers were more informal than the native
RAs, in general. The fact that features of a formal writing style, i.e. formal connectors
such as hence, therefore and nevertheless, were also prevalent in the non-native texts,
rather indicates a shift or inconsistency in register by German writers. This ’registermixing’ (Lorenz, 1999) can be related to the fact that non-native writers lack the implicit
knowledge of how to use linguistic devices appropriately in their L2. They therefore seem to
choose connectors more haphazardly than native writers facilitating the use of expressions
uncommon for the register of academic writing.
5.3.3 British and American Preferences
The study further revealed substantial differences between British and American preferences for certain connectors. Some of these preferences have already been observed by
other corpus studies. Biber et al. (1999), for example, report that as more often occurs as
a reason subordinator and while as a temporal subordinator in British English, whereas as
is more frequently used to express time or manner and while to mark contrast/concession
in American English. In the present study, as was used twice as often as a reason subordinator by British writers, and there were no instances of contrastive/concessive while in
the British RAs at all. However, the crucial point here is that many of the preferences
of the American writers are also shared by the German writers. In the present study,
reason clauses were preferably introduced by since in the German and American RAs,
while British writers clearly preferred to initiate reason clauses with as (occurring twice
as often in the British as in the American RAs). Concession, apart from however which
was frequent in all RAs, was primarily indicated by while in both German and American
English, while it was mainly introduced by although in the British RAs (exceeding the
use of although in the German and American RAs each more than fivefold). Also, more
than half of the connectors that were used by German and by either British or American
scientists were found to be used by American writers.
5. Summary of Findings and Discussion
64
The fact that German connector usage more closely resembles that of American scientists suggests that German scientific writers rather follow American than British writing
conventions. This can be attributed to the greater international influence of the American
variety. American English has a much greater impact on foreign language use than the
British variety due to its omni-presence in the media and its increasing dominance in business, communication, science and technology. It has also a dominant role in the academic
world, especially with regard to academic publishing (Hyland, 2006, see also section 2.3).
Related to this, it is interesting to note that academic papers by British and American
writers have been found to be different in terms of formality, i.e. American papers are
more informal and colloquial than British ones (cf., for example, the comparison of British
and American written genres by Biber, 1987). The finding that German EFL writing is
marked by a more informal use of connectors thus again could be due to a greater impact of
the more colloquial American conventions of English academic writing on foreign language
writers. It seems more likely, however, that the greater use of informal connectors in the
German RAs resulted from a lack of sensitivity to register as already noted above.
5.4 Connector Position: A Pointer to Non-Native Pattern?
A significant difference was found with regard to connector position. Sentence connectors
were much more frequently employed in sentence-initial position by non-native writers (over
40 % of causal and temporal connectors were used sentence-initially by German scientists
as compared to a percentage of less than 25 % in the native RAs; see table 4.13). Native
writers, in contrast, made frequent use of inter-clausal positions, including combinations
of conjunctions and linking adverbials such as and so, and but also (cf. table 4.6). While
the initial position, according to Biber et al. (1999), is the unmarked position for linking
adverbials, common connectors such as thus, therefore and however also often occur in
medial positions in academic texts. The positioning of connectors thereby by no means
appears to be at random but rather constitutes a difference in meaning (as shown in section
3.4). Medial positions not only allow for greater variation in sentence structure, but also
“point the reader more firmly to content” (Field and Yip, 1992, p. 26). It therefore seems
to call for more complex and mature writing abilities to insert connectors within sentences
than placing them at the beginning.
In the present corpus, the sentence-initial use of therefore was found to be markedly nonnative as almost three fourth of its occurrences appeared in the beginning of a sentence.
5. Summary of Findings and Discussion
65
In the native English RAs it was scarcely used sentence-initially, but occurred primarily
in non-initial position, i.e. immediately after the subject or modal verb, or in combination
with the coordinator and. The same was true for the position of so which is considered
rather informal when used initially. The greater use of initial position in non-native writing
suggests that L2 writers feel less competent to integrate connectors into sentences. The
German writers thus seemed to have avoided positions they were not sure of and rather
preferred the unmarked initial position which they felt to be less demanding in terms of
grammatical complexity. This again corresponds to what is known as avoidance strategy
in L2 research, and, in this case, resulted in an overrepresentation of the initial position in
the non-native texts (cf. section 2.3.1 for an overview on L2 acquisition theories).
Another particularly non-native usage was the use of conjunctions in sentence-initial
position. The corpus analysis showed that the German EFL writers used sentence-initial
but rather frequently while native speakers of English seemed to have certain reservations
about it (cf. table 4.5). It has also been noted that sentence-initial coordinators are considered inappropriate in formal academic writing (cf. section 3.4) which would explain
the reluctance of the English scientists to employ the coordinator but in the beginning
of a sentence. Kies (1994), however, already observed in 1993 that but is undergoing a
’semantic reanalysis’ acting more like a conjunctive adverb than a coordinating conjunction when used sentence-initially. Thus, unlike ’true’ coordinators, the ’adverbial but’ no
longer conjoins subordinate clauses or constituents smaller than a clause and it does not
necessarily link constituents of equal rank. Semantically, the adverbial but tends to imply
concession whereas the coordinator but typically signals contrast or negation. Example
(32) illustrates the use of sentence-initial but as found in one of the non-native RAs:
(32) Within this analytical concept the dependence of the intensive on the extensive parameters
has only been derived for the tip position of piezoelectric bending beam structures. The
matrix elements combining the intensive and extensive parameters have not been formulated
in a closed form as a function of any point x over the entire length of the beam actuator yet.
But just this aspect adds important insights into piezoelectric design. (GE06Bal)
In this example, the adverbial but links the sentence it initiates to the whole preceding
paragraph (instead of coordinating equal syntactic structures) and, semantically, takes the
role of a concessive linking adverbial such as however. In addition, the German equivalent
aber seems to be even more privileged to be classified as a conjunctive adverb since it is
not restricted to clause-initial position as the German coordinators und (’and’) and oder
5. Summary of Findings and Discussion
66
(’or’), or but in English. Instead, the German coordinating conjunction aber can also occur
within the clauses and constituents it links together as illustrated in the following example
given by Pasch et al. (2003, p. 683) (note the position of but in (33b) which is syntactically
correct in the German example but incorrect in the English translation):
(33a) Ich bin krank, aber ich gehe zur Arbeit. (I am ill, but I go to work.)
(33b) Ich bin krank, ich gehe aber zur Arbeit. (*I am ill, I go but to work.)
It is for this reason that in their Handbuch der deutschen Konnektoren (Handbook of German Connectors) Pasch et al. (2003) prefer to classify aber (’but’) as a special kind of adverbial connector rather than a coordinating conjunction. The remarkable use of sentenceinitial but in the German EFL writing thus might be induced by the more ’adverbial-like’
character of the German equivalent aber involving a more casual attitude towards beginning a sentence with but by German writers. Whether this is really due to L1 interference
or rather to a restricted sensitivity to register can only be established, however, by closer investigation, for example, by examining the use and positioning of aber in German academic
texts. Nevertheless, the fact that native speakers of English avoid using sentence-initial
but in scientific writing still raises the question as to the formality and appropriateness of
initial coordinators in specific registers (see section 3.4 for a more detailed discussion on
that).
67
6 Conclusion and Implications for Future
Research
The present study was designed to examine sentence connectors in native and non-native
English academic writing. For this purpose, a 73,000 word corpus was established comprising 34 English research articles by German and native (British and American) English
scientists. The comparison of the research articles revealed notable differences between
German and native English writers in their use of sentence connectors. Although, due to
the limited number of research articles, some of these differences might rather be determined by the authors’ individual styles and preferences, there are important tendencies
that emerged from the results of the study.
The choice and position of connective expressions, in particular, were found to be prone
to a markedly non-native style. As the study has shown, German scientific writers tend to
prefer the unmarked initial sentence position for connectors that could equally be placed
within the sentence. They tend to overuse expressions they are particularly familiar with,
but also opt for connectors that are more characteristic of informal registers than academic
writing. The sentence-initial use of conjunctions, i.e. of the coordinator but, appears to
be especially distinctive for German writers. English scientific articles by German writers
further tend to be marked by linguistic features that are more typical of less mature writing.
This includes the choice of because instead of as or in that, but also involves a remarkable
use of initial instead of inter-clausal positions.
The research articles of the German scientists thus reflect some typically non-native patterns that have also been observed for speakers of English as a second or foreign language
from other cultural backgrounds. These non-native patterns, such as the use of connectors
of an inappropriate register, clearly point to certain limits in L2 proficiency. The use of
less mature forms and structures, however, also suggests that the German writers were
perfectly aware of certain deficiencies in their L2 competence and therefore preferred linguistic devices they were sure of or felt more comfortable with. This learner strategy can
6. Conclusion and Implications for Future Research
68
in fact compensate for L2 deficiencies in that typical sources of misuse are avoided.
In addition, the comparison of sentence connectors gave some indication for an influence
of German L1 writing patterns. Although the informal use of connectors was found to be
characteristically non-native in general, in the case of sentence-initial but the special status
of the German equivalent aber seems to suggest an interference from L1. Also, the lower
number of connectors in the German RAs complied with findings that academic German
tends to be less explicit than academic English. Though sentence connectors are markers of
explicitness, however, “textual explicitness cannot be directly equated with explicitness of
propositional content” (Mauranen, 1993, p. 259). The connection between the number of
explicit devices and the more indirect writing practices of German writers should therefore
be subject to further consideration.
The study moreover revealed striking similarities between German and American writers
in their use of connectors in contrast to British connector usage. As, in general, it is
the American variety that is considered the norm in international academic discourse,
this seems to be a rather pleasant result for it indicates that German scientific writers
of English as a foreign language tend to conform to international standards of academic
English. Nevertheless, it remains questionable as to whether it is “desirable for language
learners to emulate the verbal behavior of native speakers” (Lorenz, 1999, p. 72). The
growing number of non-native English speakers in international communication has led
to a decreasing impact of native speaker patterns and to an increasing tolerance towards
non-native norms of English. Non-nativeness thus may be less detrimental for EFL writers
than generally assumed. On the other hand, well-developed writing skills also involve the
ability to write idiomatically, that is, to use structures and expressions that do not sound
’un-English’ to proficient writers of English. Idiomaticity is closely connected to nativeness
in that it “hinges on what kinds of structures are most likely to occur” (ibid). It therefore
seems reasonable for non-native writers to adhere to native English standards.
One pedagogical consequence of this is that academics using English as a foreign language need to be informed about existing patterns of non-nativeness and need to be given
alternatives as used by native speakers of English. This also means that linguistic research
must continue to explore the differences between native and non-native written texts in
order to determine patterns of native and non-native language use more closely. It is
important that non-native writers understand where the deficiencies lie in their L2 production so that they can systematically improve their writing performance. Furthermore,
6. Conclusion and Implications for Future Research
69
they need practice and training in the skillful use of linguistic devices. With regard to
sentence connectors, this includes that non-native writers need to become acquainted with
the different linguistic registers and the terms and expressions which are appropriate for
it. They also need to become familiar with English syntax and word order choices to gain
certainty in the utilization of connector positions besides the beginning of the sentence.
Additionally, EFL teaching should aim at strengthening non-native writers’ awareness of
culture-specific writing practices to reduce L1 interference where necessary.
To conclude, the paper in hand has yielded useful insights into the differences and similarities of native and non-native connector usage. At the same time, however, it has raised
some issues that need to receive further consideration. Thus, there was only limited scope
to investigate the reasons and consequences of certain non-native features more precisely.
It is therefore suggested to conduct a further study comparing the findings of the present
analysis with the use of connectors in the German research articles of German scientists.
This would allow a closer look into the influences of German writing practices on the foreign language production of German writers. An analysis of the English teaching material
German scholars are exposed to during school and higher education could further reveal
useful information about the impact of formal EFL instruction onto the use of cohesive
devices in German EFL compositions.
In addition, the present study has focused solely on sentence connectors, which are
one of many means to create cohesion and coherence. Cohesion is established by various
linguistic devices, such as reference, substitution and reiteration, and coherence is achieved
by more elusive features, such as the global structure and thematic progression in texts.
The corpus established for this study provides a basis to compare the use of various other
cohesive or rhetorical features between native and non-native English academic writers.
The consideration of those features in conjunction would then offer a comprehensive picture
of the textual quality of non-native scientific writing.
viii
Primary Sources
Research Articles - Native English
(BE86Bot) Bott, B. and Jones, T. A. (1986). The use of multisensor systems in monitoring hazardous atmospheres. Sensors and Actuators, 9, pages 19-25.
(AE96Bro) Brody J. P. et al. (1996). A planar microfabricated fluid filter. Sensors and
Actuators A: Physical, 54, pages 704-708.
(AE06Dar) Darabi, J. and Rhodes, C. (2006). CFD modeling of an ion-drag micropump.
Sensors and Actuators A: Physical, 127, pages 94-103.
(AE86Ell)
Elliott, D. J. and Piwczyk, B. P. (1986). Electronic materials surface processing with excimer lasers. Microelectronic Engineering, 5, pages 435-444.
(BE86Eva) Evans, N. J., Petty, M. C. and Roberts, G. G. (1986). Interface state effects
in Pd-gate mos hydrogen sensors. Sensors and Actuators, 9, pages 165-175.
(AE96Fie)
Field, L. A. et al. (1996). Micromachined 1 x 2 optical-fiber switch. Sensors
and Actuators A: Physical, 53 (1-3), pages 311-315.
(BE06Har) Harris, N. R. et al. (2006). A multilayer thick-film PZT actuator for MEMs
applications. Sensors and Actuators A: Physical, 13, pages 311-316.
(AE86Has) Haslam, M. E. and McDonald J. F. (1986). An image fidelity approach to
measuring the point spread function in electron and ion beam lithographies.
Microelectronic Engineering, 5, pages 491-498.
(BE96Hat) Hatfield, J. V. et al. (1996). An integrated multichannel charged-particle
sensing system. Sensors and Actuators A: Physical, 54, pages 777-781.
(BE06Her) Herbertson D. L. et al. (2006). Electrowetting on superhydrophobic SU-8
patterned surfaces. Sensors and Actuators A: Physical, 130-131, pages 189193.
(BE86Jon) Jones, G. A. C., Blythe, S. and Ahmed, H. (1986). Direct fabrication of
nanometre-scale structures in semiconductors with 500 keV lithography. Microelectronic Engineering, 5, pages 265-271.
Primary Sources
ix
(AE06Mal) Maltabes, J. G. and Mackay, R. S. (2006). Current overview of commercially
available imprint templates and directions for future development. Microelectronic Engineering, 83, pages 933-935.
(AE06Nor) Northen, M. T. and Turner, K. L. (2006). Meso-scale adhesion testing of integrated micro- and nano-scale structures. Sensors and Actuators A: Physical,
130-131, pages 583-587.
(AE06Pea) Pease, R. F. (2006). Direct, in-scanner, aerial image sensing. Microelectronic
Engineering, 83, pages 1030-1035.
(AE96Spa) Sparks, D. R., Jordan, L. and Frazee, J. H. (1996). Flexible vacuumpackaging method for resonating micromachines. Sensors and Actuators A:
Physical, 55, pages 179-183.
(BE96Wel) Welham, C. J., Gardner, J. W. and Greenwood, J. (1996). A laterally driven
micromachined resonant pressure sensor. Sensors and Actuators A: Physical,
52, pages 86-91.
(BE96Wil) Williams, C. B. and Yates, R. B. (1996). Analysis of a micro-electric generator for Microsystems. Sensors and Actuators A: Physical, 52, pages 8-11.
Research Articles - German EFL
(GE06Ans) Ansorge, E. et al. (2006). Piezoelectric driven resonant beam array in langasite (La3Ga5SiO14). Sensors and Actuators A: Physical, 132 (1), pages
271-277.
(GE06Bal) Ballas, R. G., Schlaak H. F. and Schmid, A. J. (2006). The constituent
equations of piezoelectric multilayer bending actuators in closed analytical
form and experimental results. Sensors and Actuators A: Physical, 130-131,
pages 91-98.
(GE96Ber) Bergner, H., Hempel, K. and Sargsjan, G. (1996). Contactless detection
of optical induced current in microelectronic devices by capacitive coupling.
Microelectronic Engineering, 31, pages 115-122.
(GE86Bie)
Biermann, M. (1986). Elastic trunnion as three-dimensional force and torque
transducer. Sensors and Actuators, 9, pages 277-286.
(GE86Ehm) Ehm, H. and Laubmeier, R. (1986). Fully-automated optical inspection for
VLSI production. Microelectronic Engineering, 5, pages 555-562.
(GE86Her) Hersener, J. and Lässing, G. (1986). A MOSFET, manufactured with synchrotron X-ray lithography. Microelectronic Engineering, 5, pages 105-112.
Primary Sources
x
(GE96Löc) Löchel, B. et al (1996). Microcoils fabricated by UV depth lithography and
galvanoplating. Sensors and Actuators A: Physical, 54, pages 663-668.
(GE96Mac) Mack, S., Baumann, H. and Gösele, U. (1996). Gas development at the interface of directly bonded silicon wafers: investigation on silicon-based pressure
sensors. Sensors and Actuators A: Physical, 56, pages 273-277.
(GE86MüS) Müller, K. H., Stelter, T., Ponse, F. and Weidlich, H. (1986). Synchrotron
lithography for sub-half-micron T-Gates in GAAS-FET. Microelectronic Engineering, 5, pages 239-246.
(GE86MüL) Müller, R. and Lange, E. (1986). Multidimensional sensor for gas analysis.
Sensors and Actuators, 9, pages 39-48.
(GE96Oes) Oesterschulze, E. and Stopka, M. (1996). Imaging of thermal properties
and topography by combined scanning thermal and scanning tunneling microscopy. Microelectronic Engineering, 31, pages 241-248.
(GE06Reu) Reuter, D. et al. (2006). In-process gap reduction of capacitive transducers.
Sensors and Actuators A: Physical, 126, pages 211-217.
(GE06Ric)
Ricken, W., Schoenekess, H. C. and Becker, W.-J. (2006). Improved multisensor for force measurement of pre-stressed steel cables by means of the eddy
current technique. Sensors and Actuators A: Physical, 129, pages 80-85.
(GE96Ros) Rossberg, D. (1996). Optical properties of the integrated infrared sensor.
Sensors and Actuators A: Physical, 54, pages 793-797.
(GE06Sch) Schultes, G. et al. (2006). Strain sensitivity of TiB2, TiSi2, TaSi2 and WSi2
thin films as possible candidates for high temperature strain gauges. Sensors
and Actuators A: Physical, 126, pages 287-291.
(GE96Sel)
Seltmann, R. et al. (1996). New system for fast submicron optical direct
writing. Microelectronic Engineering, 30, pages 123-127.
(GE06Ton) Tonisch, K. et al. (2006). Piezoelectric properties of polycrystalline AlN thin
films for MEMS application. Sensors and Actuators A: Physical, 132, pages
658-663.
xi
Secondary Sources
Ahrens, R., Bald, W.-D., and Hüllen, W., editors (1995). Handbuch: Englisch als Fremdsprache (HEF). Erich Schmidt.
Armstrong, J. S. (2007). Significance tests harm progress in forecasting. International
Journal of Forecasting, 23 (2), pages 321-327.
Biber, D. (1987). A textual comparison of British and American writing. American Speech,
62, pages 99-119.
Biber, D. (1988). Variation Across Speech and Writing. Cambridge University Press.
Biber, D. (1995). Dimensions of Register Variation: A Cross-Linguistic Comparison.
Cambridge University Press.
Biber, D. (2006). University Language: A Corpus-based Study of Spoken and Written
Registers. John Benjamins.
Biber, D., Conrad, S., and Reppen, R. (1998). Corpus Linguistics: Investigating Language
Structure and Use. Cambridge University Press.
Biber, D., Johansson, S., Leech, G., Conrad, S., and Finegan, E. (1999). Longman Grammar of Spoken and Written English. Longman.
Bourne, J. (1998). English for speakers of other languages. In Mercer, N. and Swann, J.,
editors, Learning English: Development and Diversity, pages 243–282. Routledge.
Bublitz, W. and Lenk, U. (1999). Disturbed coherence: ‘fill me in’. In Bublitz, W., Lenk,
U., and Ventola, E., editors, Coherence in Spoken and Written Discourse. How to Create
it and how to Describe it, pages 153–174. John Benjamins.
Celce-Murcia, M. and Larsen-Freeman, D. (1983). The Grammar Book: An ESL/EFL
Teacher’s Course. Heinle & Heinle.
Chang, Y. and Swales, J. M. (1999). Informal elements in English academic writing:
Threats or opportunities for advanced non-native speakers? In Candlin, C. N. and
Hyland, K., editors, Writing: Texts, Processes and Practices, pages 145–167. Longman.
Clyne, M. (1987). Cultural differences in the organization of academic texts: English and
German. Journal of Pragmatics, 11, pages 211-247.
Secondary Sources
xii
Clyne, M. (1991). The sociocultural dimension: The dilemma of the German-speaking
scholar. In Schröder, H., editor, Subject-oriented Texts: Languages for Special Purposes
and Text., pages 49–67. de Gruyter.
Connor, U. (1996). Contrastive Rhetoric: Cross-Cultural Aspects of Second-Language Writing. Cambridge University Press.
DeBeaugrande, R. and Dressler, W. (1981). Introduction to Textlinguistics. Longman.
Ellis, R. (1992). Understanding Second Language Acquisition. Oxford University Press.
Evensen, L. S. (1990). Pointers to superstructure in student writing. In Connor, U. and
Johns, A. M., editors, Coherence in Writing: Research and Pedagogical Perspectives,
pages 169–183. TESOL.
Field, Y. and Yip, L. M. O. (1992). A comparison of internal conjunctive cohesion in
the English essay writing of Cantonese speakers and native speakers of English. RELC
Journal of Pragmatics, 23, pages 15-28.
Goodwin, P. (2007). Should we be using significance tests in forecasting research? International Journal of Forecasting, 23, pages 333–334.
Grabe, W. and Kaplan, R. B. (1996). Theory and Practice of Writing: An Applied Linguistic Perspective. Longman.
Greenbaum, S. and Quirk, R. (1990). A Student’s Grammar of the English Language.
Longman.
Halliday, M. A. K. (2004). The Language of Science. Continuum.
Halliday, M. A. K. and Hasan, R. (1976). Cohesion in English. Longman.
Hinds, J. (1990). Inductive, deductive, quasi-inductive: Expository writing in Japanese,
Korean, Chinese and Thai. In Connor, U. and Johns, A., editors, Coherence in Writing:
Research and Pedagogical Perspectives, pages 87–109. TESOL.
Hinkel, E. (2002). Second Language Writers’ Text: Linguistic and Rhetorical Features.
Lawrence Erlbaum Associates.
Hoadley-Maidment, E. and Mercer, N. (1998). English in the academic world. In Mercer,
N. and Swann, J., editors, Learning English: Development and Diversity, pages 283–319.
Routledge.
Hopkins, E. A. (1982). Contrastive analysis, interlanguage, and the learner. In Lohnes, W.
F. W. and Hopkins, E. A., editors, The Contrastive Grammar of English and German,
pages 32–48. Karoma.
Secondary Sources
xiii
Hulková, I. (2007). What hides behind the use of connective adjuncts in academic texts:
Cohesion, coherence, or pragmatics? In Schmied, J., Haase, C., and Povolná, R., editors,
Complexity and Coherence: Approaches to Linguistic Research and Language Teaching,
pages 147–170. Cuvillier.
Hyland, K. (2006). English for Academic Purposes: An Advanced Resource Book. Routledge.
Kies, D. (1994). Adverbial ’but ’. In The Twentieth LACUS Forum 1993, pages 315–332.
Jupiter Press.
Lee, M. Y.-P. (2003). Structure and cohesion of English narratives by Nordic and Chinese
students. Nordlyd, 31 (2), pages 290-302.
Leech, G., Deuchar, M., and Hoogenraad, R. (2006). English Grammar for Today. Palgrave
Macmillan.
Leech, G. and Svartvik, J. (1994). A Communicative Grammar of English. Longman.
Lorenz, G. (1999). Learning to cohere: Causal links in native vs. non-native argumentative writing. In Bublitz, W., Lenk, U., and Ventola, E., editors, Coherence in Spoken
and Written Discourse. How to Create it and how to Describe it, pages 55–76. John
Benjamins.
Louwerse, M. and Graesser, A. (2005). Coherence in discourse. In Strazny, P., editor,
Encyclopedia of Linguistics, pages 216–218. Fitzroy Dearborn.
Martin, J. R. and Rose, D. (2003). Working with Discourse: Meaning Beyond the Clause.
Continuum.
Mauranen, A. (1993). Cultural Differences in Academic Rhetoric: A Textlinguistic Study.
Lang.
Milton, J. and Tsang, E. (1993). A corpus-based study of logical connectors in EFL
students’ writing: Directions for future research. In Pemberton, R. and Tsang, E.,
editors, Studies in Lexis, pages 215–246. HKUST.
Müller, S. (2005). Discourse Markers in Native and Non-native English Discourse. John
Benjamins.
Pasch, R., Brauße, U., Breindl, E., and Waßner, U. H. (2003). Handbuch der deutschen
Konnektoren. de Gruyter.
Raimes, A. (2005). Keys for Writers. Houghton Mifflin.
Reid, J. (1992). A computer text analysis of four cohesive devices in English discourse by
native and non-native writers. Journal of Second Language Writing, 1 (2), pages 79-107.
Secondary Sources
xiv
Rudolph, E. (1984). The role of conjunctions and particles for text connexity. In Conte, M.E., Petöfi, J. S., and Sözer, E., editors, Text and Discourse Connectedness, Proceedings
of the Conference on Connexity and Coherence, Urbino, July 16-21, 1984, pages 175–190.
John Benjamins.
Rudolph, E. (1996). Contrast: Adversative and Concessive Relations and their Expressions
in English, German, Spanish, Portuguese on Sentence and Text Level. De Gruyter.
Schmied, J. (2007). Complexity and coherence in English student writing, especially in
hypertext learning systems. In Schmied, J., Haase, C., and Povolná, R., editors, Complexity and Coherence: Approaches to Linguistic Research and Language Teaching, pages
13–30. Cuvillier.
Schwanzer, V. (1991). Syntaktisch-stilisitische Universalia in den wissenschaftlichen
Fachsprachen. In Bungarten, T., editor, Wissenschaftssprache: Beiträge zur Methodologie, theoretischen Fundierung und Deskription, pages 213–230. Wilhelm Fink.
Seidelhofer, B. (2005). Key concepts: English as a lingua franca. ELT Journal, 59 (4),
pages 339-341.
Silva, T. (1993). Toward an understanding of the distinct nature of L2 writing: The ESL
research and its implications. TESOL Quarterly, 27 (4), pages 657-677.
Sinclair, J., editor (1990). Collins COBUILD English Grammar. Collins.
Swales, J. M. (1990). English in Academic Research Settings. Cambridge University Press.
Swales, J. M. and Feak, C. B. (2004). Academic Writing for Graduate Students. Essential
Tasks and Skills. 2nd Edition. University of Michigan Press.
Ventola, E. and Mauranen, A. (1991). Non-native writing and native revising of scientific
articles. In Ventola, E., editor, Functional and Systematic Linguistics: Approaches and
Uses, pages 457–492. Mouton de Gruyter.
Wikborg, E. (1990). Types of coherence breaks in Swedish student writing: Misleading
paragraph division. In Connor, U. and Johns, A., editors, Coherence in Writing: Research and Pedagogical Perspectives, pages 131–149. TESOL.
Witte, S. and Faigley, L. (1981). Coherence, cohesion and writing quality. College Composition and Communication, 32, pages 189-204.
xv
Appendix A - Additional Data on Primary
Material
In the following, some detailed information is given about the research articles that were
used as primary sources of this study. The tables contain information about the authors
and their academic degree or position (as far as specified), the institutional context as well
as the publishing journal, title, language and word count of the RAs. The word count
only refers to the RAs’ continuous text and does not include abstracts, acknowledgments,
legends, foot- or endnotes, and other additional texts.
Abbreviations
ME - Journal Microelectronic Engineering, Elsevier
SA - Journal Sensors and Actuators (A: Physical), Elsevier
n.s. - not specified
Research Articles - Native English
published 2006 (Table A.1)
published 1996 (Table A.2)
published 1986 (Table A.3)
Research Articles - German EFL
published 2006 (Table A.4)
published 1996 (Table A.5)
published 1986 (Table A.6)
Journal
SA
ME
SA
ME
SA
SA
Code
AE06Dar
AE06Mal
AE06Nor
AE06Pea
BE06Har
BE06Her
University of South Carolina, Columbia,
SC, USA
Photronics Inc., Austin, TX, USA
Professor;
2 Professor
4 Professor
1 PhD;
2 PhD
3 Professor
Townsend3b ;
Beeby1a ; White4a
Herbertson1 ;
Evans2 ;
Shirtcliffe1 ;
Newton1
3 PhD
Torah1a ;
McHale3 ;
1 PhD; 2 n.s.;
Harris1a ; Hill2b ;
student;
student;
n.s.
nano-scale structures”
b Mechanical
Professor
Pease
integrated micro- and
California, Santa Barbara, CA, USA;
2 Associate
Nottingham, UK
Sciences, Nottingham Trent University,
of Southampton, UK
School of Biomedical and Natural
surfaces”
hydrophobic SU-8 patterned
“Electrowetting on super-
applications”
b School
of Engineering Sciences, University
actuator for MEMs
Science, University of Southampton, UK;
Total
BE
BE
sensing"
“A multilayer thick-film PZT
a School
of Electronics and Computer
AE
AE
"Direct, in-scanner, aerial image
future development”
“Meso-scale adhesion testing of
AE
AE
Language
Santa Barbara, CA, USA
Stanford University, Stanford, CA, USA
Engineering Dep., University of California,
and Environmental
Department, University of
Turner2b
a Materials
1 PhD
templates and directions for
commercially available imprint
Northen1a ;
“Current overview of
micropump”
“CFD modeling of an ion-drag
Title
n.s.
student;
Department of Mechanical Engineering,
Institution
1 Assistant
Degree/Position
Academic
Maltabes; Mackay
Darabi1 ; Rhodes2
Author(s)
Table A.1: Research Articles - Native English published 2006
12637
1999
2120
1994
2200
1611
2713
Words
A. Additional Data on Primary Material
xvi
Journal
SA
SA
SA
SA
SA
SA
Code
AE96Bro
AE96Fie
AE96Spa
BE96Hat
BE96Wel
BE96Wil
Research Lab.;
Greenwood3b
Williams1 ; Yates2
2 Director
Gardner2a ;
student;
2 M.Sc.
1 PhD
3 B.Sc.
of Sensors
student;
1 PhD
Welham1a ;
Director
4 Managing
Comer4b
student;
resonant pressure sensor”
Engineering, Sheffield, UK
Department of Electronic and Electrical
University of Sheffield, MEMS Unit,
Groby, Leicester, UK
generator for Microsystems”
“Analysis of a micro-electric
“A laterally driven micromachined
a Department
of Warwick, Coventry, UK; b Druck Ltd,
charged-particle sensing system”
Manchester, UK
of Engineering, University
UK; b Integrated Sensors Limited,
and Electronics, UMIST, Manchester,
3 PhD
Hicks2a ;
Lomas3b ;
micromachines”
“An integrated multichannel
a Department
1 PhD; 2 Professor;
Hatfield1a ;
of Electrical Engineering
method for resonating
switch”
“Flexible vakuum-packaging
“Micromachined 1 x 2 optical-fiber
USA
CA, USA
Delco Electronics Corp., Kokomo, IN,
filter”
“A planar microfabricated fluid
Title
Frazee3
1 PhD; 2 M.S.; 3 B.S.
Robrish1 ; Ruby1
Sparks1 ; Jordan2 ;
Washington, Seattle, WA, USA
Hewlett-Packard Laboratories, Palo Alto,
1 PhD; 2 n.s.
4 Professor
Field1 ; Burriesci2 ;
of Mechanical Engineering, University of
Professor;
Forster3b ; Yager4a
for Bioengineering, University of
Washington, Seattle, WA; b Department
2 PhD; 3 Associate
a Center
Osborn2a ;
Professor;
Institution
1 Assistant
Degree/Position
Academic
Brody1a ;
Author(s)
Table A.2: Research Articles - Native English published 1996
Total
BE
BE
BE
AE
AE
AE
Language
12912
2032
2210
2317
2132
1889
2332
Words
A. Additional Data on Primary Material
xvii
Journal
ME
ME
SA
SA
ME
Code
AE86Ell
AE86Has
BE86Bot
BE86Eva
BE86Jon
Ahmed
Jones; Blythe;
n.s.
1 PhD; 2 Professor
Evans1 ; Petty2 ;
Roberts1
n.s.
n.s.
n.s.
Degree/Position
Academic
Bott; Jones
McDonald
Haslam;
Elliott; Piwczyk
Author(s)
semiconductors with 500 keV
University, UK
lithography”
nanometre-scale structures in
“Direct fabrication of
Cavendish Laboratory, Cambridge
Durham, UK
Microelectronics Research Group,
“Interface state effects in Pd-gate
Sheffield, UK
Department of Applied Physics and
mos hydrogen sensors“
monitoring hazardous atmospheres”
and Laboratory Services Division,
Electronics, University of Durham,
lithographies”
“The use of multisensor systems in
function in electron and ion beam
Health and Safety Executive, Research
measuring the point spread
NY, USA
processing with excimer lasers”
“An image fidelity approach to
Rensselaer Polytechnic Institute, Troy,
“Electronic materials surface
Leitz-Image Micro Systems Co., Billerica,
Title
MA, USA
Center for Integrated Electronics,
Institution
Table A.3: Research Articles - Native English published 1986
Total
BE
BE
BE
AE
AE
Language
10977
1955
2664
2198
2102
2058
Words
A. Additional Data on Primary Material
xviii
Journal
SA
SA
SA
SA
SA
SA
Code
GE06Ans
GE06Bal
GE06Reu
GE06Ric
GE06Sch
GE06Ton
Schmidt3a
3 Professor
Billep2a ; Scheibner2a ;
Magdeburg,
University of Technology,
“In-process gap reduction of
Germany
a Center for Microtechnologies, Chemnitz
IZM, Micro Devices and
a Institute
1 PhD
2 Assistant
Tonisch1a ; Cimalla2a ;
Foerster3a ;
Ambacher3a ;
Dontsov3b
3 PhD
temperature strain gauges”
“Piezoelectric properties of
Sensortechnology, Saarbrücken, Germany
Romanus3a ;
as possible candidates for high
Department of Mechatronics and
2 Dipl.Ing.; 3 PhD
Göttel2 ; Freitag-Weber3
Professor;
student;
TiSi2, TaSi2 and WSi2 thin films
Hochschule für Technik und Wirtschaft,
1 Professor;
Meßtechnik GmbH, Ilmenau, Germany
Technical University Ilmenau, Germany;
b SIOS
for Micro- and Nanotechnologies,
MEMS application”
Total
polycrystalline AlN thin films for
eddy current technique”
“Strain sensitivity of TiB2,
steel cables by means of the
Engineering, Kassel, Germany
Schultes1 ; Schmitt2 ;
measurement of pre-stressed
Engineering/Computer Sciences, Measurement
3 Professor
“Improved multi-sensor for force
1 PhD; 2 Dipl.Ing.;
Becker3
Equipment, Chemnitz, Germany
University of Kassel, Department of Electrical
b Fraunhofer
capacitive transducers”
form and experimental results”
Technology & Testing, Esslingen-Berkheim,
University of Technology, Chemnitz, Germany;
actuators in closed analytical
Germany; b Festo AG & Co. KG, Department
“The constituent equations of
piezoelectric multilayer bending
University of Technology, Institute
(La3Ga5SiO14)”
beam array in langasite
“Piezoelectric driven resonant
Title
for Electromechanical Design, Darmstadt,
a Darmstadt
Germany
Institute of Metallurgy, Clausthal-Zellerfeld,
Germany;
b Clausthal
Institute of Micro and Sensor Systems (IMOS),
a Otto-von-Guericke-University
Institution
Ricken1 ; Schoenekess2 ;
Dötzel3a ; Gessner3ab
1 Dipl.-Ing., 2 PhD
1 PhD; 2 Professor
Reuter1a ; Bertz2a ;
Schmid1b
Ballas1a ; Schlaak2a ;
Fritze3b ;
2 Dipl.-Ing.; 3 PhD
student;
1 PhD
Sauerwald2b ;
Hirsch1a ;
Degree/Position
Academic
Ansorge1a ; Schimpf1a ;
Author(s)
Table A.4: Research Articles - German EFL published 2006
13010
2136
2288
2201
1968
2243
2174
Words
A. Additional Data on Primary Material
xix
Journal
ME
SA
SA
ME
SA
ME
Code
GE96Ber
GE96Löc
GE96Mac
GE96Oes
GE96Ros
GE96Sel
Pauflera ; Zimmerab
Germany
b Duisburg,
“New system for fast submicron
Germany
a Fraunhofer Institute of Microelectronic
Kücka ; Melcherb ;
infrared sensor”
Technische Universität München, Munich,
Total
optical direct writing”
microscopy”
“Optical properties of the integrated
Lehrstuhl für Technische Elektronik,
Circuits and Systems, Dresden, Germany;
n.s.
Seltmanna ;
topography by combined scanning
Kassel, Germany
thermal and scanning tunneling
pressure sensors”
“Imaging of thermal properties and
investigation on silicon-based
directly bonded silicon wafers:
lithography and galvanoplating”
“Gas development at the interface of
devices by capacitive coupling”
“Microcoils fabricated by UV depth
induced current in microelectronic
“Contactless detection of optical
Title
Institute of Technical Physics, University of
Bosch GmbH, Reutlingen, Germany
Physics, Halle (Saale), Germany;
b Robert
Berlin, Germany
a Max-Planck-lnstitute of Microstructure
Doleschala ; Gehnera ;
n.s.
Rossberg
n.s.
2 PhD; 3 Professor
Gösele3a
Oesterschulze; Stopka
1 PhD
Mack1a ; Baumann2b ;
student;
3 n.s.
Rothe3 ;
Windbracke1
Fraunhofer-Institut für Siliziumtechnologie,
1 PhD; 2 Dipl.Ing.;
Jena, Germany; b Friedrich
Löchel1 ; Maciossek2 ;
a Fachhochschule
Institution
Schiller University Jena, Germany
n.s.
Degree/Position
Academic
Sargsjanb
Bergnera ; Hempelb ;
Author(s)
Table A.5: Research Articles - German EFL published 1996
13419
1749
2146
2246
2241
2477
2560
Words
A. Additional Data on Primary Material
xx
Journal
SA
ME
ME
SA
ME
Code
GE86Bie
GE86Ehm
GE86Her
GE86MüL
GE86MüS
n.s.
Weidlichb
Müllera ; Steltera ;
Ponseb ;
n.s.
n.s.
Hersenera ; Lessingb
Müller; Lange
n.s.
Dipl.Ing.
Degree/Position
Academic
Ehm; Laubmeier
Biermann
Author(s)
analysis”
“Synchrotron lithography for
electronic, Heilbronn, FRG
Lehrstuhl für Technische Elektronik,
Technische Universität München, Munich,
FRG
a Siemens Systemtechnik, Berlin, FRG;
GAAS-FET”
Total
sub-half-micron T-Gates in
synchrotron X-ray lithography”
“Multidimensional sensor for gas
b TELEFUNKEN
Components Division, Munich, FRG
for VLSI production”
“A MOSFET, manufactured with
Microelectronics, Munich, FRG
a AEG Research Center, Ulm, FRG;
b Siemens
transducer”
“Fully-automated optical inspection
three-dimensional force and torque
Berlin, FRG
Siemens A.G. Technology Center for
“Elastic trunnion as
Title
Federal Institute for Material Testing (BAM),
Institution
Table A.6: Research Articles - German EFL published 1986
10112
1871
2011
2043
2112
2075
Words
A. Additional Data on Primary Material
xxi
xxii
Appendix B - Additional Tables
Table B.1: Sentence Connectors Examined
Additive
Adversative
additionally
also
as well
besides
for example / e.g.
for instance
further
furthermore
in addition
indeed
in fact
likewise
moreover
similarly
that is / i.e.
too
and
or
Causal
(A) Linking Adverbials
conversely
as a result
however
consequently
in contrast
for this reason
instead
hence
nevertheless
otherwise
nonetheless
so
on the contrary
then1
on the other hand therefore
yet
thus
Temporal
afterwards
finally
first
firstly
first of all
next
now
second
secondly
then2
thirdly
(B) Coordinating Conjunctions
but
(C) Subordinating Conjunctions
although
as1
even though
as long as
though
because
whereas
if
while1
in order + to
in that
since
(so) as + to
so that
unless
whether
after
as2
before
when
while2
xxiii
B. Additional Tables
Table B.2: Statistical Significance of Differently Normalized Connector Frequencies
Word Basis
1,000
10,000
100,000
NCF* Native English
NCF* German EFL
24.91
21.73
249.14
217.29
2491.40
2172.90
2.197
21.682
χ2
0.191
*Normalized connector frequencies
Table B.3: Distribution across Syntactic Categories (items per 10,000 words)
LinkAdv
CConj
SConj
AE
BE
Native English
German EFL
83.55 (37.15 %)
77.24 (34.34 %)
64.11 (28.51 %)
85.74 (31.12 %)
110.32 (40.04 %)
79.45 (28.84 %)
84.60 (33.96 %)
93.08 (37.36 %)
71.46 (28.68 %)
84.29 (38.79 %)
69.51 (31.99 %)
63.49 (29.22 %)
Table B.4: Clause-level Conjunctions (items per 10,000 words)
AE
BE
Native English
German EFL
Clause-level and
All occurrences
Clause-level but
All occurrences
Clause-level or
All occurrences
Clause-level yet
All occurrences
64.63
236.98
8.41
10.51
4.20
28.37
-
96.03
241.21
12.58
15.43
1.71
31.44
-
79.67
239.01
10.40
12.87
3.01
29.84
-
56.10
241.92
11.49
15.87
1.64
22.17
0.27
0.27
Clause-level Total
CConj Total
77.24
275.87
110.32
288.08
93.08
281.72
69.51
280.23
Table B.5: Distribution across Semantic Categories (items per 10,000 words)
Additive
Adversative
Causal
Temporal
AE
BE
Native English
German EFL
100.89 (44.86 %)
26.27 (11.68 %)
52.02 (23.13 %)
45.71 (20.32 %)
120.03 (43.57 %)
34.87 (12.66 %)
76.59 (27.80 %)
44.01 (15.97 %)
110.06 (44.18 %)
30.39 (12.20 %)
63.79 (25.60 %)
44.90 (18.02 %)
91.68
27.91
71.70
26.00
(42.19
(12.84
(33.00
(11.97
%)
%)
%)
%)
xxiv
B. Additional Tables
Table B.6: Additive Connectors (items per 10,000 words)
AE
additionally
initial
also
initial
as well
initial
besides
initial
for example / e.g.
initial
for instance
initial
further
initial
furthermore
initial
in addition
initial
indeed
initial
in fact
initial
likewise
initial
moreover
initial
similarly
initial
that is / i.e.
initial
too
initial
and
initial
or
initial
Total
initial
BE
Native English
(A) Linking Adverbials
3.68
1.92
3.68
1.92
17.34
10.29
13.96
1.58
0.57
1.10
0.53
0.27
3.16
2.29
2.74
0.53
2.29
1.37
0.57
0.27
0.53
0.57
0.55
1.58
0.82
1.58
0.82
2.63
3.43
3.01
2.63
2.86
2.74
1.05
0.55
0.53
0.27
0.57
0.27
0.57
0.27
1.14
0.55
1.14
0.55
0.57
0.27
0.57
0.27
1.58
2.29
1.92
0.57
0.27
(B) Coordinating Conjunctions
64.63
96.03
79.67
0.53
0.27
4.20
1.71
3.01
100.91
11.03
120.02
8.00
110.06
9.58
German EFL
1.09
0.82
16.15
1.37
1.09
0.27
0.27
4.65
0.27
0.27
0.27
0.55
0.27
3.83
3.56
0.82
0.82
0.27
0.27
0.27
0.27
0.27
0.55
0.55
1.64
2.19
56.10
1.64
91.67
8.76
xxv
B. Additional Tables
Table B.7: Adversative Connectors (items per 10,000 words)
AE
conversely
initial
however
initial
in contrast
initial
instead
initial
nevertheless
initial
nonetheless
initial
on the contrary
initial
on the other hand
initial
(A)
8.93
5.78
0.53
0.53
-
BE
Native English
Linking Adverbials
0.57
0.27
0.57
0.27
12.00
10.40
8.57
7.12
0.27
0.27
0.57
0.27
0.57
0.27
-
German EFL
4.10
3.01
0.27
0.27
0.27
1.09
1.09
0.55
0.27
1.09
0.82
but
initial
yet
initial
(B) Coordinating Conjunctions
8.41 12.58
10.40
-
11.49
3.28
0.27
-
although
initial
even though
initial
though
initial
whereas
initial
while 1
initial
(C) Subordinating Conjunctions
1.05
7.43
4.11
1.05
2.29
1.64
0.53
0.27
0.53
1.71
1.10
0.53
0.27
6.31
3.29
3.15
1.64
1.37
1.09
0.55
3.01
0.82
3.83
0.82
Total
initial
26.27
11.03
34.87
12.00
30.39
11.50
27.91
11.49
xxvi
B. Additional Tables
Table B.8: Causal Connectors (items per 10,000 words)
AE
as
initial
as long as
initial
because
initial
if
initial
in that
initial
in order + to
initial
since
initial
so as + to
initial
so that
initial
unless
initial
whether
initial
Total
initial
Native English
German EFL
(A) Linking Adverbials
1.58
0.82
1.05
0.55
1.14
0.55
1.14
0.55
0.53
0.27
0.53
0.27
1.05
3.43
2.19
1.14
0.55
2.63
5.72
4.11
0.53
0.27
3.68
3.43
3.56
1.58 10.86
6.02
1.71
0.82
6.31
8.57
7.39
3.68
3.43
3.56
0.27
0.27
0.27
0.55
0.55
2.19
1.37
1.92
0.55
3.28
1.64
5.47
0.55
10.67
7.94
5.75
2.46
(C) Subordinating Conjunctions
3.68
8.00
5.75
1.71
0.82
0.53
0.27
2.10
3.43
2.74
0.57
0.27
7.88
6.86
7.39
2.63
4.57
3.56
1.58
1.14
1.37
5.25
8.57
6.84
1.05
2.29
1.64
6.83
2.29
4.65
4.73
0.57
2.74
1.05
0.57
0.82
4.73
9.15
6.84
0.53
0.57
0.55
0.53
2.86
1.64
-
1.92
1.37
0.27
0.27
5.20
16.69
8.21
6.84
2.46
6.29
3.56
0.55
2.46
1.09
-
as a result
initial
consequently
initial
for this/these reason/s
initial
hence
initial
otherwise
initial
so
initial
then1
initial
therefore
initial
thus
initial
1
BE
52.02
13.14
76.59
17.15
63.79
15.61
71.70
31.20
xxvii
B. Additional Tables
Table B.9: Temporal Connectors (items per 10,000 words)
AE
BE
Native English
(A) Linking
1.71
1.14
2.29
1.71
0.57
0.57
2.29
1.14
0.57
0.57
9.15
0.57
0.57
German EFL
Adverbial
2.46
1.92
4.11
1.64
0.55
0.55
1.10
0.55
0.27
1.10
0.82
11.22
0.27
0.27
afterwards
initial
finally
initial
first
initial
firstly
initial
first of all
initial
next
initial
now
initial
second
initial
secondly
initial
then2
initial
thirdly
initial
3.15
2.63
5.78
1.58
0.53
0.53
0.53
1.58
1.05
13.14
-
after
initial
as2
initial
before
initial
when
initial
while2
initial
(C) Subordinating Conjunctions
2.10
4.00
3.01
0.53
0.57
0.55
6.83
4.57
5.75
2.10
0.57
1.37
1.58
2.29
1.92
0.57
0.27
7.36 14.86
10.95
1.58
4.57
3.01
3.15
1.14
2.19
-
4.38
3.01
0.27
1.37
0.27
5.75
1.37
1.64
0.55
Total
initial
45.71
9.98
26.00
11.49
44.01
12.00
44.90
10.95
0.82
0.55
1.37
1.09
2.74
1.09
0.82
0.82
0.55
0.27
3.01
1.09
0.27
0.27
0.55
2.46
1.09
-
xxviii
Zusammenfassung
Motivation und Zielstellung
Die vorliegende Arbeit befasst sich mit der Verwendung von Satzkonnektoren in englischsprachigen wissenschaftlichen Publikationen englischer Muttersprachler und Nichtmuttersprachler. Satzkonnektoren sind Verbindungswörter, die einzelne Sätze oder Textpassagen
zu einem zusammenhängenden Text verbinden. Sie spielen eine wichtige Rolle bei der formalen und inhaltlichen Strukturierung von Texten und tragen wesentlich zur Kohärenz,
dem inneren Zusammenhalt von Texten bei. Da gerade Fachtexte ein hohes Maß an Klarheit
und Strukturierung erfordern, stellen Satzkonnektoren ein wichtiges sprachliches Element
bei der Vermittlung wissenschaftlicher Inhalte dar.
Unter der zunehmenden Dominanz des Englischen als Wissenschaftssprache sehen sich
auch nichtenglischsprachige Akademiker immer häufiger dazu veranlasst, ihre wissenschaftlichen Erkenntnisse und Innovationen auf Englisch zu veröffentlichen. Dies erfordert nicht
nur ein hohes Maß an Fremdsprachenkompetenz, sondern ebenso Kenntnisse über die Stilund Schreibkonventionen der englischen Wissenschaftssprache. Textlinguistische Untersuchungen in kontrastiver Rhetorik weisen deutlich darauf hin, dass Textsorten (sog. Genre)
und Schreibstile kulturspezifisch sind (vgl. z. B. Connor, 1996; Grabe and Kaplan, 1996).
Dies gilt auch für Fachtextsorten wie wissenschaftliche Publikationen. Des Weiteren haben vergleichende Studien in der Fremdsprachenforschung gezeigt, dass muttersprachliche
und nicht-muttersprachliche Texte in wesentlichen Merkmalen voneinander abweichen (vgl.
z. B. Hinkel, 2002). Dies betrifft die Makrostruktur von Texten (z. B. den Aufbau und die
Anordnung größerer funktionaler Einheiten wie Kapitel und Paragraphen) als auch die
Verwendung spezifischer lexikalischer und grammatikalischer Einheiten, wie Pronomen,
Partikel oder Satzkonnektoren. So haben die Sprachwissenschaftler auch herausgefunden,
dass es nichtmuttersprachlichen englischen Texten oftmals an Kohäsion (dem formalen Zusammenhalt) und Kohärenz (dem inhaltlichen Zusammenhalt) fehlt. Die Gründe hierfür
lassen sich zum einen auf die kulturspezifischen Stil- und Schreibkonventionen der jeweili-
Zusammenfassung
xxix
gen Nichtmuttersprachler zurückführen (vgl. Hinds, 1990; Mauranen, 1993), zum anderen
auf einen defizitären oder missbräuchlichen Einsatz kohäsiver Mittel wie Referenzpronomen
und Verbindungswörter (vgl. u.a. Hinkel, 2002; Milton and Tsang, 1993). Englischsprachige
Fachtexte, die nicht den Konventionen der englischen Wissenschaftssprache genügen, werden jedoch von Muttersprachlern oft als unstrukturiert und inkohärent empfunden. Dies
kann für nichtenglischsprachige Akademiker von großem Nachteil sein, wirkt sich die Art
und Weise wie Fachwissen in einem Text vermittelt wird doch auch auf die Beurteilung
fachlicher Kompetenz aus.
Die vorliegende Studie trägt dazu bei, Unterschiede und Gemeinsamkeiten in der Verwendung von Satzkonnektoren zwischen englischen Muttersprachlern und Nichtmuttersprachlern aufzudecken. Insbesondere sollen die Häufigkeit, Vielfalt und Stellung der Satzkonnektoren in englischensprachigen Fachtexten deutscher und britischer bzw. amerikanischer
Autoren verglichen werden. Dafür wurde ein Textkorpus erstellt, der 34 wissenschaftliche Publikationen (sog. „research articles“) mit insgesamt 73.000 Wörtern umfasst, die
zu gleichen Teilen von deutsch- und englischsprachigen Wissenschaftlern im Bereich Physik und Elektrotechnik auf Englisch verfasst und in den Jahren 1986, 1996 und 2006 in
den Fachzeitschriften Microelectronic Engineering und Sensors and Actuators (A: Physical) veröffentlicht wurden. Die Korpustexte gehören der gleichen Textsorte an und erfüllen
die gleiche Textfunktion: die Vermittlung von Fachwissen. Außerdem wurden sie bewusst
aus der gleichen wissenschaftlichen Disziplin ausgewählt, um sicher zu stellen, dass sie gemeinsamen Stil- und Schreibkonventionen entsprechen. Sie stellen somit eine einheitliche
Grundlage dar, um die Verwendung spezifischer sprachlicher Mittel in muttersprachlichem
und nichtmuttersprachlichem Englisch zu untersuchen.
Der Studie liegen folgende Untersuchungsfragen zugrunde:
1. Verwenden deutsche Autoren in englischen Publikationen Satzkonnektoren in gleichem Maße wie Muttersprachler (Häufigkeit der Konnektoren)?
2. Wählen deutsche Autoren die gleichen englischen Satzkonnektoren und die gleiche
Vielfalt wie Muttersprachler (Vielfalt an Konnektoren)?
3. Verwenden deutsche Autoren englische Satzkonnektoren in den gleichen Positionen
(z. B. satzinitial vs. satzmedial) wie Muttersprachler (Stellung der Konnektoren)?
Des Weiteren sollen die eben genannten Dimensionen Häufigkeit, Vielfalt und Stellung der
Konnektoren jeweils für die Jahre 1986, 1996 und 2006 betrachtet werden um folgende
Zusammenfassung
xxx
Frage zu klären:
4. Hat sich die Verwendung von Satzkonnektoren in muttersprachlichen und nichtmuttersprachlichen englischen Fachtexten in den letzten 20 Jahren verändert?
Schließlich soll bei der Betrachtung der muttersprachlichen Fachtexte zwischen britischem
und amerikanischem Englisch unterschieden werden um herauszufinden:
5. Inwieweit weichen deutsche Autoren in ihrer Verwendung von Satzkonnektoren von
Muttersprachlern des amerikanischen bzw. britischen Englisch ab?
Methodik
Zur Klasse der Satzkonnektoren zählen Adverbkonnektoren („linking adverbials“) wie therefore und however sowie konjunktionale Konnektoren, also Konjunktoren („coordinating
conjunctions“) und Subjunktoren („subordinating conjunctions“) wie and, but und because. Mit Hilfe des Konkordanzprogramms AntConc3.2.1w wurde der Korpus zunächst nach
Satzkonnektoren aus den genannten syntaktischen Klassen durchsucht und die Anzahl ihres Vorkommens sowie ihre Stellung im Satz für jeden der 34 Texte einzeln dokumentiert.
Wörter und Ausdrücke, die mehrere syntaktische Funktionen erfüllen können, wurden unter gesonderter Betrachtung ihrer Verwendung im Kontext manuell bestimmt. So können
beispielsweise die Subjunktoren after, as, since und before als Konnektoren, aber auch
als Prepositionen fungieren. Außerdem wurden nur Verbindungswörter berücksichtigt, die
vollständige Haupt- und Nebensätze, z. B. Adverbialsätze, miteinander verknüpfen. Konjunktoren, die der Verbindung einzelner Wörter oder Satzteile, wie z. B. Nominal- und
Prepositionalphrasen, dienen, wurden nicht zur Kategorie der Satzkonnektoren gezählt.
Zusätzlich zur syntaktischen Klassifizierung wurden die Satzkonnektoren entsprechend
ihrer semantischen Funktion differenziert. Die Einteilung in semantische Klassen erfolgte nach den von Halliday and Hasan (1976) vorgegebenen Bedeutungsrelationen additiv,
adversativ, kausal und temporal sowie den dazugehörigen Subrelationen. Satzkonnektoren, die mehrere Bedeutungsrelationen gleichzeitig ausdrücken können, wie beispielsweise
as (kausal vs. temporal) und while (adversativ vs. temporal), wurden entsprechend ihrer Verwendung im Text der jeweils naheliegensten semantischen Kategorie zugeordnet.
Um schließlich einen direkten Vergleich der Verwendung der einzelnen Satzkonnektoren zu
gewährleisten, wurden die jeweiligen Häufigkeitsraten auf die Anzahl der Konnektoren je
10.000 Wörter normiert.
Zusammenfassung
xxxi
Ergebnisse und Auswertung
Die Ergebnisse der Korpusanalyse lassen sich entsprechend den oben gestellten Untersuchungsfragen wie folgt zusammenfassen:
1. Häufigkeit: Deutsche Autoren verwenden weniger Satzkonnektoren in ihren englischsprachigen Publikationen als englische Muttersprachler.
2. Vielfalt: Deutsche Autoren verwenden eine größere Vielfalt an englischen Satzkonnektoren. Dazu gehören auch informelle Konnektoren sowie Konnektoren, die häufig
in sprachlich weniger ausgereiften Texten vorkommen.
3. Stellung: Deutsche Autoren verwenden Satzkonnektoren erheblich häufiger am Satzanfang (satzinitial) als englische Muttersprachler.
Die Betrachtung der einzelnen Jahrgänge ergab folgendes Ergebnis:
4. Über die Jahre 1986, 1996 und 2006 blieb die Anzahl der verwendeten Satzkonnektoren in den muttersprachlichen Publikationen weitestgehend konstant, während sie
in den Artikeln der deutschen Wissenschaftler größeren Schwankungen ausgesetzt
war. Unterschiede in der Wahl und Vielfalt der Satzkonnektoren blieben im gesamten Zeitraum gleich. Der häufige Gebrauch von Satzkonnektoren am Satzanfang in
den nichtmuttersprachlichen Texten hat hingegen drastisch zugenommen. Eine Annäherung des muttersprachlichen und nichtmuttersprachlichen Gebrauchs von Satzkonnektoren war nicht zu beobachten.
Die Differenzierung von britischem und amerikanischem Englisch erbrachte eine weitere
entscheidende Erkenntnis:
5. Der Gebrauch von Satzkonnektoren in englischen Publikationen deutscher Autoren
ist dem der amerikanischen Muttersprachler deutlich ähnlicher als dem der britischen
Muttersprachler. Dies zeigt sich in der insgesamten wie auch den einzelnen Häufigkeiten ebenso wie in Präferenzen für bestimmte Satzkonnektoren.
Die Unterschiede im Gebrauch von Satzkonnektoren zwischen Muttersprachlern und
Nichtmuttersprachlern der englischen Sprache können auf verschiedene Ursachen zurückgeführt werden. So legt die insgesamt geringere Verwendung von Konnektoren in den Publikationen der deutschen Autoren einen gewissen Einfluss der deutschen Muttersprache
nahe. Kulturvergleichende Studien haben gezeigt, dass die deutsche Wissenschaftssprache
Zusammenfassung
xxxii
im Vergleich zur englischen weitaus indirekter und weniger explizit ist (Clyne, 1987, 1991).
Satzkonnektoren sind Ausdruck eines expliziten Schreibstils, da sie die semantischen Relationen zwischen Propositionen klar und deutlich hervorheben. Ein Einfluss der impliziten
Ausdrucksweise des Deutschen ist daher wahrscheinlich. Ebenso muss jedoch ein zustande kommen der Unterschiede aufgrund individuell verschiedener Schreibstile der Autoren
berücksichtigt werden.
Abweichungen in der Vielfalt und Auswahl spezifischer Konnektoren lassen hingegen die
unterschiedlich ausgeprägten Sprachkompetenzen von Muttersprachlern und Nichtmuttersprachlern als Ursache vermuten. Die Studie hat gezeigt, dass sich die Wahl der Satzkonnektoren in den von deutschen Autoren verfassten Publikationen eher willkürlich gestaltet,
während Muttersprachler gezielt auf bestimmte Konnektoren zurückgreifen. Dies begründet
auch das erhöhte Vorkommen informeller Konnektoren wie besides, too und as well sowie
den Gebrauch von so und but am Satzanfang in den nichtmuttersprachlichen Texten. Die
Verwendung informeller Konnektoren ist ein oft beobachtetes Phänomen in fremdsprachlichen Texten und wird auf ein mangelndes Bewusstsein für einen angemessenen Stil bzw.
das passende Register zurückgeführt (vgl. z. B. Celce-Murcia and Larsen-Freeman, 1983).
Einige Satzkonnektoren stellten sich auch als sogenannte „lexical teddy-bears“ heraus, also
Wörter und Ausdrücke mit denen die Autoren besonders vertraut sind und aus diesem
Grund überproportional häufig verwendet werden. Die Bevorzugung des kausalen Konnektors because gegenüber dem aufgrund seiner syntaktischen und semantischen Ambiguität
anspruchsvollerem as seitens der deutschsprachigen Autoren weist auf eine bekannte Lernerstrategie hin, d. h. die Vermeidung von Formen und Strukturen mit deren Umgang der
Lerner sich ungeübt und unsicher fühlt. Dies erklärt auch, dass die Nichtmuttersprachler
insgesamt häufiger auf sprachlich weniger komplexe und anspruchsvolle Ausdrücke zurückgegriffen haben.
Damit verbunden ist die Tatsache, dass Satzkonnektoren in den nichtmuttersprachlichen Texten erheblich häufiger satzinitial verwendet wurden als in den muttersprachlichen
Texten. So positionierten die deutschen Autoren 40 % der kausalen und temporalen Konnektoren am Satzanfang während dies nur bei 25 % derselben in den muttersprachlichen
Publikationen der Fall war. Da die Integration von Konnektoren innerhalb des Satzes ein
komplexeres Verständnis syntaktischer Strukturen erfordert als die „unmarkierte“ Stellung
am Satzanfang, deutet dies erneut auf eine typische Vermeidungsstrategie hin.
Zusammenfassung
xxxiii
Die Differenzierung zwischen britischem und amerikanischem Englisch hat zudem ergeben, dass viele der deutschen Präferenzen für bestimmte Konnektoren (z. B. since und
while an Stelle von as und whereas) von den amerikanischen, nicht aber von den britischen Autoren geteilt werden. Auch die Häufigkeitsverteilung der Konnektoren in den
syntaktischen und semantischen Kategorien weist Ähnlichkeiten im deutschen und amerikanischen Gebrauch von Satzkonnektoren auf, während Unterschiede zwischen den mutterund nichtmuttersprachlichen Texten zumeist auf die britische Varietät zurückzuführen sind.
Deutsche Akademiker und Benutzer des Englischen als Fremdsprache scheinen demnach
der Dominanz des amerikanischen Englisch im internationalen Wissenschaftsdiskurs zu
unterliegen.
Schlussfolgerung
Die komparative Analyse englischer Publikationen deutsch- und englischsprachiger Autoren
hat deutliche Unterschiede zwischen Muttersprachlern und Nichtmuttersprachlern im Umgang mit Satzkonnektoren aufgezeigt. Abweichungen ergaben sich insbesondere hinsichtlich
muttersprachlicher und nichtmuttersprachlicher Präferenzen für bestimmte Konnektoren
sowie deren Stellung im Satz. Die Ergebnisse weisen darauf hin, dass deutsche Wissenschaftler mit der Verwendung englischer Satzkonnektoren zwar vertraut sind, aber dennoch gewisse Defizite aufweisen, die für Fremdsprachentexte jedoch nicht untypisch sind.
Etwaige Einflüsse der Muttersprache auf die Fremdsprachenproduktion konnten ebenfalls
beobachtet werden.
Die Studie kommt zu dem Schluß, dass deutsche Autoren englischer Wissenschaftstexte in
der Verwendung von Satzkonnektoren typischen nichtmuttersprachlichen Mustern folgen.
Um einen geübten und effizienten Umgang mit Satzkonnektoren zu gewährleisten, müssen
fremdsprachige Autoren auf jene Muster nichtmuttersprachlichen Sprachgebrauchs aufmerksam gemacht und über Alternativen, wie sie von Muttersprachlern verwendet werden,
aufgeklärt werden. Auch sollte ihr Bewusstsein für die verschiedenen sprachlichen Register
sowie für die jeweiligen kulturspezifischen Schreib- und Stilkonventionen verstärkt geschult
werden, um den Gebrauch unpassender Ausdrücke und Strukturen in der Zielsprache zu
vermeiden.
xxxiv
Selbstständigkeitserklärung
Ich versichere, dass ich die vorliegende Arbeit selbst verfasst und dafür ausschliesslich
die angegebenen Texte und Quellen verwendet habe. Alle wörtlich übernommenen Aussagen sind als Zitate eindeutig gekennzeichnet. Die Herkunft der indirekt übernommenen
Formulierungen ist angegeben.
Chemnitz, den 7. Juli 2008
Unterschrift des Verfassers